Sunday, September 17, 2006

Could 2007 be a Repeat of 2002 for AMD? Part II.

In part I we looked at the overall events of AMD's revenue drop in 2002. In part II we'll look at the current financial outlook of both companies.

We saw that AMD's very large revenue drop to 5.5% was only a temporary event and that the actual shift was from about 9.3% to 8.2%. However, during this time, AMD was much less profitable than Intel. There are many reasons for this. Intel had been well established in servers since the Pentium Pro while AMD did not begin to make server chips until the Athlon MP in late 2001. Intel also was more involved with mobile, manufactured its own chipsets, and obtained lower costs from its 300mm FAB. Today, AMD has a good track record and established base of server chips, is more diversified into mobile, has an operating 300mm FAB (FAB 36), and with ATI's purchase is moving into chipsets. AMD is much more diversified than it was in 2002 and this tends to make AMD more profitable.

Intel was also ahead in the shift to smaller manufacturing processes. Intel did have some difficulty in following AMD to copper interconnects which led to some of problems in 2001 however they were arguably ahead on process technology at the 130nm level. AMD had had a process partnership with Motorola and then one with UMC but it found a true partner in IBM. IBM's process experience finally enabled AMD to begin manufacturing the K8. Although Intel is still moving to smaller processes more quickly it no longer has more advanced process technology. However, even if AMD's process is slightly more complex the two processes are still very much alike. The first real test of process leadership won't occur until Intel begins using high K materials for gates while AMD stays with silicon dioxide. AMD has a small lead in process control software with APM. It takes Intel two months to move a chip from test to production with its Copy Exact method whereas AMD can go to production the same day with APM. This shows a fundamental difference in design style as Intel has dedicated FABs (D1D and D1C) specifically for design while AMD runs tests in its production FABs. We saw in the graphs that when one company had production difficulty the other company temporarily benefited however there is currently no indication of problems from either company.

There is no doubt that Conroe is faster on the desktop than X2. Even the fastest FX-62 processor is only a bit faster than the E6600. If AMD were to release a 3.0Ghz FX-64 its speed would be just slightly less than an E6700 but at a much greater price tag. Assuming that the 4X4 plan covers X6800, AMD would need to price match:

E6700 - no current match, would require 3.0Ghz clock, 5800+
E6600 - no current match, would require 2.8Ghz clock, 5400+, similar to FX-62
***** - 5000+ AMD's current fastest common chip, slower than E6600 but priced higher
E6400 - 4600+, currently matches in price
E6300 - 4200+, currently matches in price
***** - 3800+ should be comparable to a proposed lower model by Intel

AMD might skip E6700 since it would need a 3.0Ghz chip but it should probably address E6600. This would still require producing a 2.8Ghz 5400+. AMD will either have to address this or leave this upper segment entirely to Intel along with its higher margins.

Overall, things are not as bad for AMD as it would appear. In servers, the Woodcrest version of Core 2 Duo doesn't share the same advantages that the desktop chip has. This is true for several reasons. Opterons are capable of up to 8-way configurations but 4-way is currently the point of maximum benefit. Woodcrest is only capable of 2-way which leaves the 4-way market very open to AMD. Intel's older Prescott based Xeon 4-way chips are much less competitive with both lower performance and higher power draw. Further, Intel's newest 5000 chipset is designed for FBDIMM which has three disadvantages. FBDIMMs draw more power than regular RDIMMs which nullifies Woodcrest's lower power draw. FBDIMM also slows down substantially when the memory slots are fully populated as they normally are with servers. This tends to nullify Woodcrest's higher speed. Finally, Intel announced a very large scale back to its expected use of FBDIMM in 2008 while AMD canceled its plans for FBDIMM entirely. This could make Intel's 5000 series chipsets much less desirable. Intel could solve all of the problems except the 2-way limitation by releasing a new chipset for Woodcrest but this will take time. These factors tend to remove any immediate advantages that Woodcrest may have had. In the mobile area Merom has much less of a lead because the fastest Conroe chips draw more power. However, the 65nm Merom chips are ahead AMD's 90nm Turion chips. AMD should be reasonably competitive again once Turion is moved down to the 65nm process.

It is clear that for dual core Intel will be fastest no matter what in 2006. However, this is not so clear when we move up from dual core. Intel will release the Kentsfield quad core equivalent of Conroe later this year and probably the Clovertown quad core version of Woodcrest in 2007. These chips however are far less ambitious and leading edge than Conroe. In fact, they are essentially the same strategy as Intel used to lackluster effect with Smithfield. If Smithfield is any indication, Kentsfield is destined to be the quad core version of Celeron by end of 2007. Smithfield was an MCM which means just putting two dies in the same package rather than a truly integrated design like Core 2, Core 2 Duo, and K8. Further, Smithfield was manufactured on the same process which led to high power draw. Kentsfield seems to have all of the same disadvantages. Kentsfield's power draw basically doubles compared to Conroe while using two dies in the same package means that two FSB controllers share the same bus. Although a desktop unit can tolerate a higher power draw for a performance chip this factor tends to make Clovertown much worse than Woodcrest. It remains to be seen if anyone will elect to upgrade from Woodcrest to Clovertown on large scale servers at the cost of doubling the cpu power draw. This is important as was shown by the upgrade of an older, Opteron based supercomputer at Oakridge National Labs. It could do this because AMD managed to stay within its TDP when moving to dual core and has stated that it will again with quad core. This could be why Opteron now has contracts for two TeraFlops supercomputers. Intel seems to be similarly capable of doing this with integrated designs (like Yonah (Core 2) and Woodcrest) but not with MCM designs. In this light, Kentsfield and Clovertown have to be seen as stopgap until Intel creates an integrated quad core design.

So, if Intel's quad core offerings are stopgap then what is AMD's 4X4? To find out if 4X4 is a clever new idea or just a stopgap reaction to Conroe and Kentsfield we have to see how it compares to current dual socket systems. You can pick up a pair of 1.8Ghz Opteron dual cores for $560. Since only the lowest of the 4X4 processors was stated to be under $1,000 this indicates that 4X4 is a more upscale offering. It would appear that AMD is trying to span two markets with 4X4. Essentially, we have two chips replacing what was one FX chip at the same price range. In this light, it would be an enthusiast offering and match up with companies like Alienware and Voodoo PC. This seems consistent as Alienware has already stated support. So, dual core FX replaced single core and now dual chip is replacing single chip. However, making use of all four cores means either having a game that has that much threading (which none currently do) or running things in the background while playing a game. Perhaps in view of this trend, multithreading will accelerate. The other area that may be spanned by 4X4 would be workstations. However, this is mostly a question of what memory is supported as RDIMMs are required to create workstation or server class systems. If 4X4 can successfully span both enthusiast and workstation markets it could be clever idea.

C2D is definitely a boost for Intel but the volume in 2006 is not high enough to drive the overall market. By year's end, C2D will only be at 30% of Intel's capacity which is what AMD's 65nm production will be. Simply shrinking the die to 65nm and reducing power consumption should be sufficient in 2006 for the server and notebook markets. It has been suggested that Intel will take back share by lowering prices. If 4X4 will replace one chip with two then it appears that AMD is at least somewhat prepared to match Intel in terms of price. However, Intel currently has a large overstock of Prescott and Presler based P4 chips. Releasing Conroe at low prices does put pressure on AMD but it puts even more pressure on Intel's P4 line which will still make up around 80% of the total chip volume in the last two quarters. Since this will dominate Intel's revenues in 2006 I would expect Intel to see around a 12% drop versus 2005 while I would expect AMD's revenues to increase around 10% versus 2005. To put this another way, Conroe simply will not be produced in large enough volumes to turn the market around for Intel. It will bring new revenues but these will not be enough to offset the P4 declines.

Things do look different in 2007. As C2D continues to ramp up this will move things in Intel's favor and at some point C2D will begin to dominate Intel's sales. This is where it gets more interesting. Intel's ramp of C2D would be very favorable by the end of the first quarter. The only other thing in AMD's pocket to counter would be K8L, AMD's fully integrated quad core design. This chip includes a shared L3 cache, double the prefetch size, double the L1 cache bus width, and double the FP pipelines. It boosts its FP performance similar to the way C2D was boosted over Yonah. This chip has advanced power management and additional AMD64 instructions. And, because it is manufactured on 65nm it draws no more power than AMD's current dual core 90nm chips. The interesting part is that AMD officially says mid year 2007 for release. However, this estimate is similar to the current end of year estimate for 65nm chips. In both cases, AMD allows itself some padding so that if problems occur it can still meet its estimated delivery. If things go well, it can always move its release forward. This is why it seems likely that 65nm will appear in October rather than in December. Likewise, K8L could arrive in the 2nd quarter of 2007. K8L is likely to surpass Conroe in FP/SSE performance at the same clock and probably cut the current integer performance gap in half. With higher SSE performance but lower Integer performance I would say that the dual core version of K8L (without shared L3) will be about equal to Conroe. And, Intel will need a better chipset to prevent Woodcrest from being left behind. This won't actually have much effect on notebooks since K8L won't be used for mobile but it will outclass Intel's stopgap quad core offerings just as X2 did with Smithfield and the Presler die shrink. Overall, K8L should deliver more performance and lower power draw than Kentsfield/Clovertown. However, Intel is likely to respond with an integrated quad core design in 2008 as it did with C2D.

It isn't just K8L though that would make AMD competitive in 2007. Whereas Intel will be almost entirely converted to 65nm by end of 2006, AMD will only be at 30% conversion on one FAB. This means that AMD will continue to gain from lower cost as it ramps 65nm production toward mid year in 2007. Similarly, Intel will be at full 300mm production whereas FAB36 will only be at 50%. AMD again gains from lower costs as FAB36 reaches 100% capacity by end of 2007 and 125% (with a cleanroom addition currently under construction) by mid 2008. Intel's biggest benefit in 2007 will be conversion to the Conroe die. Conroe's die is only about half the size of the Presler die which will lower costs for Intel. However, the die size for K8 is actually smaller than Conroe which means that AMD will get a little more benefit. AMD's falling costs due to 300mm and 65nm ramping in 2007 puts it into a very good position to resist pricing pressures from Intel in 2007. This means that the argument for gaining share by reducing prices will essentially be gone by year's end as AMD's costs will drop faster than Intel's during 2007.

There are some incidental areas like Intel is currently taking losses in non-computing areas and is spending down its cash reserves. AMD's cash reserves are depleted and it incurs some debt with the ATI purchase. The interest on the debt is mostly offset by the cost savings of merging AMD and ATI so the debt itself is the biggest factor. However, Intel's unprofitable sections and its current large inventories are likely to hit its stock harder than AMD's. This would particularly be true if Intel has to reduce stock buyback to save money. With Dell's increased AMD offerings combined with the increased server offerings from IBM, Sun, Dell, and HP it does seem unlikely that AMD will see any large reversals soon. AMD seems to have mostly short term problems while Intel's appear to be longer term. AMD needs to address E6400 and E6600. However, AMD seems to have power management and performance under control with K8L and its 4X4 plan appears to create a new price structure in the former FX/EE range. Although Intel looks very good at the moment it will need to create an integrated quad core solution and address 4X4, HTX, and HT 3.0 in 2008 to stay competitive. Short of either Intel or AMD's having production problems that reduce volume there are unlikely to be any large changes in share. However, AMD's current 18% revenue share may just be temporary and it could fall back to 15% where it was two quarters ago. Overall, I would expect Intel to do worse this year than 2005 and AMD to do better than 2005.

43 comments:

Anonymous said...

Just a little before I go to sleep..
http://www.digitimes.com/mobos/a20060823A7037.html
http://www.intel.com/technology/architecture/coremicro/
"Clovertown is targeted for introduction in the fourth quarter of 2006, on the Bensley and Glidewell platforms."
"Taiwan makers to start delivering Clovertown-ready motherboards by end of 3Q"
And I've always thought that the general consensus was that E6600 was better than FX62. Also, their Xeon 5080/PE965/FX62 are around current Kentsfield level, so I don't see what the big deal is.

Scientia from AMDZone said...

Okay, Clovertown will be out in 2006. This is a little earlier than I had thought.

Well, I've seen some people claim that E6400 beats an FX-62. But I'm not sure what you mean by general consensus. When I wrote the article about the dishonesty of overclocking it was after having seen dishonest testing at several different websites. This is still going on. The Toms Hardware Guide review would indicate that E6600 is faster than FX-62 but this is because THG is not an honest review site. They try to tweak their setups and testing to make Intel look as good as possible. My general observation is that Conroe is 15% faster than K8 at the same clock. This makes a 2.4Ghz E6600 about the same speed as a 2.8Ghz FX-62. However, Conroe will still be faster in terms of SSE.

I'm also not sure what you mean about the big deal on Kentsfield. The current implication for AMD is that someone who buys a socket AM2 or socket F system will be able to upgrade from dual core to quad core. Kentsfield would tend to imply this same thing except power doubles for Kentsfield so I'm not sure that most motherboards will be able to handle this.

Pop Catalin Sever said...

What about prices?
Do you think it is posible that we will see a price colapse in the 'high end' somewhere between this this Cristmas and Q2007 ?
I think if 4x4 truly delivers the promised performance levels in the high end, along with the laungh of Vista and the first true multithreaded games (not dual threaded) like Crysis and DX10 (first directx API muthithreaded at the core) we might just see another price war! I'm saying this because I don't see the new Intel (the Otelini era) wanting to loose any image batles as it was shown with Conroe wich was more of a strategic marketing launch than wise moneymaking investment.

Anyways AMD is sure to win the battle once K8L is out but Intel will definately try tu turn the tables again in Q3-Q4 2007 by going maybe nativq quad core as you said and posibly improving on FP and Int performance. Unfortunatly it's still stuck on the FSB and this time just ading more cache won't have as much as impact as it did in the case of Conroe!

Unfortunately the outcome of this performance race is the widening gap between high end computing and the other two markets mainstream and low end. While the enthusiasts aremore than happy with the technology advances the software developers remain faitful to me masses. It pays more to optimize a game to run on a Sempron than to optimize it to run on a quad core, actualy it a huge diference in market posible marketshare. I sure haven't seen a game so far that doesn't run on a low end Sempron with a decent (mainstream) vid card. It's great to have all of those Cores especilay for optimized software like 3D modeling/rendering, video compresion etc... but in order to have a paradigm shift in aplication development hadware multhithreading must reach even the low end.

My point is AMD or Intel some other company could have all the leadership they want because it still doesn't make you a winner just to have technolgy leadership, you have to be able to bring it in the reach of the masses to win anything from it.

Darth Solarion said...

Hi, I was reading your blog and was wondering about the reviews you criticised as biased. Could you point out which part of the hardware setup was tweaked to showcase Intel in the best possible light? Thanks.

Scientia from AMDZone said...

No, I don't expect any dramatic change in prices at the high end, particularly if the top of the 4X4 line overlaps in workstations. This appears to maintain the previous price for FX, it's just that you would get two chips instead of one for that price.

There is no actual battle to win. AMD's goal is to continue to take share until it reaches somewhere around 30-35% volume share. This is the point where they estimate that Intel's cpu monopoly will break. That number comes from AMD's head legal counsel. It was interesting to me because it was close to the 31-37% volume share that I had come up with based on revenue share. I suppose if AMD's average selling price continues to rise, my number would move closer to his.

I don't really see a widening gap between enthusiast and desktop top end. For example, even though a 4X4 system may be expensive you could buy a dual socket motherboard and put two lower model Opteron 2xx chips on it for half price and still have a 4 core system. You could do the same with a Smithfield (or possibly Presler) dual core.

Software will change; that is inevitable. In Q1 06 only 6% of AMD's desktop chips were dual core, however, in Q4 06 46% will be dual core. This will cover 70% of the midrange so I would guess the whole midrange will be covered in Q2 07. And we should have dual core Semprons and Celerons before end of 2007.

I wouldn't really talk about technology leadership in this context. Conroe is faster than K8. Woodcrest is roughly equal to Opteron on single and dual socket systems. Turion should be closer to Merom once it moves to 65nm. What is curious is that AMD has had a single core to fill the server, desktop, and mobile roles whereas Intel has had three separate cores plus substantial differences between Xeon and P4. Now, that has reversed with AMD moving to a separate core for mobile while Intel's Woodcrest/Conroe/Merom is the same core with very few differences (and it has scaled back development of Itanium quite a bit). That Intel would copy AMD's strategy of using a single core suggests that Intel's project management is a problem.

It is also interesting that after Intel canceled the Indian designed Whitefield, AMD apparently hired most of the Whitefield design team so their pool of engineering talent should be getting very competitive with Intel's. Consider that if CSI was the major problem for Whitefield then combined with HT it would have been a complete, native quad core design. Intel still has plenty of engineering talent if they can just get project management under control.

Scientia from AMDZone said...

I was reading your blog and was wondering about the reviews you criticised as biased. Could you point out which part of the hardware setup was tweaked to showcase Intel in the best possible light?

Yes, I can do that. The only problem is that the list would be quite long. So, now I'm trying to decide if I can post this in the comments section or if I need to do a whole article.

Anonymous said...

I haven't found anyone who would say FX62 is faster than the E6600, but in gaming, the E6600 is the obvious choice, and if someone was to get FX62 to OC, I haven't seen anyone not OC their E6300/E6400 under 2.4(E6600 level).

Some people are downplaying Kentsfield because of the doubled power requirements, but don't consider that ~130W is still in the reach of those that are even considering it.

I don't think performance, wattage, or other things can be linear like that(Conroe 15% faster), though I agree, Conroe isn't that much faster(though has a lot more headroom compared to K8), since there's still a 'base' performance factor. Like even though Kentsfield is 2 E6700s, they're both sharing things(chip, FSB), so it's not exactly double performance/wattage (could be more since gas+fire=bigger fire, or less, since they're sharing)

If you're irked that Toms uses bad RAM for AMD or whatever, you could also take that as..Intel handling the memory better, and going AMD would require better memory. I'd like to see that article:)

"That Intel would copy AMD's strategy of using a single core suggests that Intel's project management is a problem."

They've been on the 'performance per watt' wagon for a while, I don't think it's a new problem, just on going change. Also, isn't AMD moving away from one core with Bulldozer?

"For example, even though a 4X4 system may be expensive you could buy a dual socket motherboard and put two lower model Opteron 2xx chips on it for half price and still have a 4 core system."

Opterons require slower, more expensive memory, expensive motherboard, and in the end, you would've spent quite a bit, for a system not meant for desktop use, might as well go Kentsfield/4x4.

I'm not familiar with server platforms, but Xeon looks like it's winning every benchmarking thrown at it except Apache(server?), but still performs better on 3D(workstation?) I think AMD will need a new core for Turion(Bulldozer). THG's benchmarks show it barely inching past Core 1 and Pentium M yet having much worse battey life. Though Turions are cheaper(which is why Intel is releasing OEM Meroms to compete)

Scientia from AMDZone said...

I haven't found anyone who would say FX62 is faster than the E6600

if someone was to get FX62 to OC, I haven't seen anyone not OC their E6300/E6400

If you're irked that Toms uses bad RAM for AMD or whatever


It will be difficult for us to have a conversation because you appear to place importance in the ability of a chip to overclock and you seem to get your ratings from Tom's Hardware Guide.

I place no importance at all in the ability of a chip to overclock. The highest estimate of people who overclock is 1 out of 2,000.

There are people who love to tinker with their cars. They enjoy spending time trying to get more performance and that is fine. The ProSport community is supported and encouraged but it is distinctly different from the vast majority of people who buy cars. The subject of overclocking does not belong in a mainstream review just as the subject of bolting on a nitrous oxide injector doesn't belong in mainstream car review.

Tom's Hardware Guide has no credibility; they gave that up many years ago before K8 was ever released. Tom's isn't the only one though and this has been going on a long time.

I can show you a 1999 XbitLabs article where the highest speed available for K7 was 700Mhz while the highest available for PIII was 600Mhz. Yet, Xbit did the comparison by using a PIII overclocked to 650 Mhz against a 600 Mhz K7. Overclocking has been used as a means of cheating in Intel's favor for a long time.

Anonymous said...

I'm sure that number is much higher among people that buy CPUs directly, especially those buying the extreme FX62 and whatnot.

There are many sites besides THG, generally showing the cooler, cheaper E6600 at least on par, many times ahead of FX62.
http://www.gamepc.com/labs/view_content.asp?id=e6600&page=1
I saw some AMD fan link to GamePC recently showing Xeon losing in Apache. Will there ever be a site that satisfies you?:) I wonder what you base your opinions of Athlon on. PS, mentioning a 7 year old article?:P

Anonymous said...

"Kentsfield would tend to imply this same thing except power doubles for Kentsfield so I'm not sure that most motherboards will be able to handle this."

These boards are also made to be able to handle PE965/PE840. Kentsfield is near this level.

http://www.hkepc.com/bbs/news.php?tid=671378&starttime=0&endtime=0

The high end quad Xeon will be 120W, mainstream are at 80W, and a low voltage at 50W.

Anonymous said...

Scientia, though I jabbed at you a little, in the post minutes before 19/9/06 5:56 PM, I believe you are too selective in what comments show, considering my last one didn't even insult you. If you care to post this one:), I said that hardware sites aren't mainstream. Anyone wanting mainstream reviews should go PCMag or something. People looking at individual retail CPUs would also be more inclined to overclock. And hardware sites in general show E6600 ahead of FX62, amazing considering that it's cheaper and cooler.

Scientia from AMDZone said...

Well, I agree that any board that can handle Presler should be able to handle Kentsfield. I'm wondering about the 5000 boards though.

Anonymous said...

http://www.xtremesystems.org/forums/showpost.php?p=1563117&postcount=29
"We are just updating all our X7xxx products to support Clovertown processors. You will have no issue if you purchase 30 days from now since we are just starting the hardware change."
Already there are benches of Clovertown on XS, one guy used a Supermicro. Anandtech tried adding Clovertowns to the Mac Pro..
http://www.anandtech.com/mac/showdoc.aspx?i=2832&p=6
"We grabbed a pair of 2.4GHz Clovertown samples and tossed them in the system, and to our pleasure, they worked just fine."
TDP of Dempsey was 95/130W, Clovertown will be 80/120W.
PS, my previous 2 posts were non derogatory, please stop censoring:)

Scientia from AMDZone said...

I'm sure that number is much higher among people that buy CPUs directly, especially those buying the extreme FX62 and whatnot.

Okay, let's say that the number of people who buy cpu's directly is 100X higher for overclocking. Then we have 1/20 instead of 1/2000. This would still only be 5%. However, you aren't understanding me. I'm not against overclocking. But, overlocking should be in separate articles from reviews. I'm sure most people are familiar with marketing. For example, products the store wants you to buy go at eye level. Products of less importance go on the top and bottom shelves. I don't want to get into all of this here because this is lengthy and I'm thinking about doing an article but if you always have an overclocked Intel cpu in a review with a normally clocked AMD cpu then that is an attempt to create an association (it's marketing and not honest).

There are many sites besides THG, generally showing the cooler, cheaper E6600 at least on par, many times ahead of FX62.
http://www.gamepc.com/labs/view_content.asp?id=e6600&page=1


Again, I don't have time here to go over benchmarking quality, site bias, testing variations, etc. I'll see if I can do an article.

I saw some AMD fan link to GamePC recently showing Xeon losing in Apache.

There are several different areas of computation on servers including HPC, transaction speed, and database searching.

Will there ever be a site that satisfies you?:) I wonder what you base your opinions of Athlon on.

Well, if I pull together a website review article then I guess you should be able to tell.

PS, mentioning a 7 year old article?:P

Well, I had thought that showing a 7 year old article went along with the statement that bias in reviews had been around a long time.

Scientia from AMDZone said...

I'm going to try this without word verification but might have to turn it back on if I start getting automated comment spam.

Scientia from AMDZone said...

Scientia, though I jabbed at you a little, in the post minutes before 19/9/06 5:56 PM,

I don't know the specific post you are referring to. If you want to associate your posts you'll something other than "anonymous".

I believe you are too selective in what comments show, considering my last one didn't even insult you.

Actually, it's not a matter of tone. I had two posts that said that my first version of this article was too long; I just posted one of those.

As I say in the description on the first page, I don't tend to publish posts that are purely complimentary or purely insulting. However, I haven't gotten any that were purely insulting. I have gotten several that I haven't published that were purely complimentary (which I appreciate). I don't publish them because I don't want any clutter in the comments and I don't want to be self promoting.

If you care to post this one:),

The interface for this blog is a bit primitive. For example, I can see what comments have been made but I can't reply to them in that window, nor can I tell what article they are from. I have to check my email notification to see what article they go with or I have to resort the comments in article order.

It is difficult to publish a comment and then reply to it later because after I publish comments they disappear from the list. Secondly, they don't show up in the order that I publish them; they show up in the order they were submitted. This means that if I don't comment when I publish I can easily forget which ones I've replied to. I do receive a notification of publishing which might be enough to tell which ones I've replied to; I might try that and publish comments sooner if I can keep track of them.

I said that hardware sites aren't mainstream. Anyone wanting mainstream reviews should go PCMag or something. People looking at individual retail CPUs would also be more inclined to overclock. And hardware sites in general show E6600 ahead of FX62, amazing considering that it's cheaper and cooler.

Yes, I commented on this one.

Scientia from AMDZone said...

http://www.xtremesystems.org/forums/showpost.php?p=1563117&postcount=29
"We are just updating all our X7xxx products to support Clovertown processors.


I've already agreed that the 7000 series which was made for Presler and Smithfield should be able to handle Kentsfield. I'm still wondering about the 5000 chipset.

Already there are benches of Clovertown on XS, one guy used a Supermicro. Anandtech tried adding Clovertowns to the Mac Pro..
http://www.anandtech.com/mac/showdoc.aspx?i=2832&p=6
"We grabbed a pair of 2.4GHz Clovertown samples and tossed them in the system, and to our pleasure, they worked just fine."
TDP of Dempsey was 95/130W, Clovertown will be 80/120W.


That only leaves the question of the 5000 server chipset and the single socket P4 boards. I assume you are saying that any board that can handle a dual core Smithfield can handle a Kentsfield but what about boards designed for single core P4's?

PS, my previous 2 posts were non derogatory, please stop censoring:)

I assume they got published. You might use a name instead of "anonymous" though for reference. I'm hoping turning off the word identification works out; it seems a bit easier.

Anonymous said...

All I have to say is that I'm sure Intel likes THG and THG likes Intel:)
It is shady how they overclock Intels more, but in the end, many other sites show basically the same thing.
http://www.planetx64.com/index.php?option=com_content&task=view&id=283&Itemid=14
AMD fan site
"They were not satisfied with meeting the AM2’s performance, they surpassed it significantly. Even with the latencies inherent in having the memory controller in the northbridge, Intel’s new architecture is able to pretty much pound AMD into submission."

Anonymous said...

Mac Pro uses 5000 I believe. Single socket P4s?
..
http://www.dailytech.com/article.aspx?newsid=4253
"Intel is planning to release quad-core Kentsfield based Xeon 3000 seriesprocessors. The new Xeon X3220 and X3210 processors will be identical to the recently named Intel Core 2 Quadro processors and share the same Kentsfield core."

Scientia from AMDZone said...

http://www.planetx64.com/index.php?option=com_content&task=view&id=283&Itemid=14
AMD fan site

"they surpassed it significantly.

Intel’s new architecture is able to pretty much pound AMD into submission"


I'm not sure what point you are trying to make. Intel did surpass AMD significantly. Comparing a 2.8Ghz chip with a 2.4Ghz chip is 2 1/2 speed grades difference. Down 1 speed grade is behind, two and half is a pounding.

Scientia from AMDZone said...

http://www.planetx64.com/index.php?option=com_content&task=view&id=283&Itemid=14
AMD fan site


BTW, I'm not sure why you mentioned that it was an AMD fan site. Perhaps you assume that because I mention AMDZone that I'm biased against Intel, etc. I've seen the name Scientia used in other places so I wanted it to be clear which one I was. I've bought both AMD and Intel based systems. I suppose if I were a diehard AMD fan I would link to Sharikou's site. Instead, I've posted corrections on his site several times. It doesn't do any good; he has blinders on.

My perspective tends to be a bit different. And, my prediction rate has been better than 90% accurate for more than three years. For example, back in 2003 when Anandtech was saying how great Prescott would be I was saying that I didn't think Intel could add on another generation; I don't know of anyone else who was saying that. I'd say I got that one right. BTW, I tend to be slightly optimistic about Intel, not pessimistic.

Anonymous said...

They said it not me, and I brought it up since you seem not content with other hardware sites, so here's an AMD site:)

And I don't understand your speed grades? But I think it makes more sense to compare CPUs by price, rather than clocks and pipelines and such.

Scientia from AMDZone said...

Single socket P4s?

Well, it seems like you should be able to put a Conroe (or a Kentsfield) in any socket 775 board. That would be an upgrade path.

Scientia from AMDZone said...

They said it not me,

Who is "they"?

and I brought it up since you seem not content with other hardware sites, so here's an AMD site:)

The quality of a site isn't determined by which processor they prefer. It's more important to be honest.

And I don't understand your speed grades?

Sorry, I didn't say that clearly.

X6800 to E6600 is 530Mhz which would be 2 1/2 speed grades (at 200Mhz each). The processor that FX should be close to is X6800, but it isn't.

But I think it makes more sense to compare CPUs by price

If you've read my article then you know that I already did. E6300 is roughly a match for 4200+ both in terms of performance and price. E6400 is roughly a match for 4600+ both in performance and price. Above E6400, AMD isn't currently competitive. AMD would need 3.0Ghz just to match E6700 and it doesn't have that.

Anonymous said...

Sean Kalinich of PlanetX64 is 'they'

Scientia from AMDZone said...

It isn't on the home page. I did a search but couldn't find the phrase anywhere on the website either.

Anonymous said...

http://www.planetx64.com/index.php?option=com_content&task=view&id=283&Itemid=14&limit=1&limitstart=9http://www.planetx64.com/index.php?option=com_content&task=view&id=283&Itemid=14&limit=1&limitstart=9

Page 10 of 10

Conclusion:
My original intention with this evaluation was to cover rendering, virtualization, and multi-tasking in depth. However the official release date was moved up by two weeks cutting my testing time significantly. I will be following up this hands-on evaluation with in-depth articles and evaluations on those topics, that (due to time constraints) had to be left out. The evaluation as it stands is still enough to show that Intel has finally pulled their collective heads out of the sand and gotten back into the performance game… and with a vengeance. They were not satisfied with meeting the AM2’s performance, they surpassed it significantly. Even with the latencies inherent in having the memory controller in the northbridge, Intel’s new architecture is able to pretty much pound AMD into submission. There will still be some areas that AMD will out perform Conroe. But they will be few and far between for now. Should you dump AMD and run to Intel? Well, again, that is something for you to decide. I am certain that AMD is not sitting back and shaking their heads. They are at work on something. Of course time will tell if that product will put them back in the game or not.

Scientia from AMDZone said...

Assuming that Kentsfield is compatible with most socket 775 and 771 boards this still leaves the original problem. You are talking about around 120 watts for a single Kentsfield versus 130 watts for two K8L's. This will cut Intel's performance/watt in half compared to AMD and also prevent Kentsfield's being used in any high density or large scale servers. However, I think Kentsfield and Clovertown would work fine in regular single and dual socket systems.

Anonymous said...

I believe AMD has said they'll increase performance yet keep the same thermal envelope, not lower it.
Woodcrest 51XXs are 40W (low voltage) 65W (mainstream) 80W(performance).
Socket F is 68W (low voltage) 95W (mainstream) 120W performance.
Socket F is a good ~50% higher than Intel TDP wise.
65nm has reduced AM2 ~27% according to the tweakers.net article.
Assuming a 27% reduction across the board(very hypothetical, since TDP gets exponential), that would make for 50W low voltage, 70W mainstream, and 88W, just for dualcore. Quads would obvioulsy be hotter, unless they're clocked really low(what Intel's going to do with the 50W quad). I expect AMD to probably clock them so that they'll match the previous generation's TDP of 68/95/120(link to where AMD said they'll have !a 68W part).



http://www.tgdaily.com/2006/08/15/amd_releases_socket_f_and_am2_opteron/

"AMD confirmed to TG Daily that it will be offering a 68W quad-core processor."
Just offering, means they'll be the lowest. 50W is less than 68W. Sounds like a really low voltage/underclocked CPU just to grab the TDP crown.

Anonymous said...

Just a summarizartion from my last post, and more direct response to your post:D

http://www.tgdaily.com/2006/08/15/amd_releases_socket_f_and_am2_opteron/
"..and consume the !same power as today's dual-core processors: AMD confirmed to TG Daily that it will be offering !a 68W quad-core processor."

Today's dual core Opteron 22XXs are currently rated 68/95/120W.
http://www.hkepc.com/bbs/news.php?tid=671378&starttime=0&endtime=0
Quad core Xeons are rated 50/80/120.

Scientia from AMDZone said...

Today's dual core Opteron 22XXs are currently rated 68/95/120W.
http://www.hkepc.com/bbs/news.php?tid=671378&starttime=0&endtime=0
Quad core Xeons are rated 50/80/120.


You have to multiply Intel's ratings by 20% to make them match AMD's ratings. AMD uses maximum power whereas Intel uses typical power. So, we get:

Xeon - 60, 96, 144
K8L - 68, 95, 120

Still pretty close.

Scientia from AMDZone said...

You have to multiply Intel's ratings by 20% to match AMD's. However, the numbers are still pretty close.

Anonymous said...

Theoretical 20%:)
But even with your numbers
Core 2 started out lower so it had more room to grow. K8L(Barcelona) must have some magical powers if it'll have 4 cores, same TDP as previous generation, and still be clocked high enough for single threaded stuff.

Scientia from AMDZone said...

Theoretical 20%:)

It's Intel's theory that if the cpu draws more than it's partial TDP number that the cpu will be saved by thermal throttling. However, it isn't fair to compare a typical max number with an actual max number.

But even with your numbers
Core 2 started out lower so it had more room to grow.


Actually, it didn't. Let's add the 15 watts for the memory controller and see what happens.

Xeon - 75, 111, 159
K8L - 68, 95, 120

K8L(Barcelona) must have some magical powers if it'll have 4 cores, same TDP as previous generation,

It isn't magic. There's actually a lot of common sense involved.

Original Opteron - 130nm

Dual core Opteron - 90nm plus limited power control.

Quad core Opteron - 65nm plus sophisticated power control.

Woodcrest - 65nm plus sophisticated power control.

Basically, the reason why K8L can stay in the same thermal range better than Woodcrest is because K8L hasn't used the advantages of the 65nm process and sophisticated power control as Woodcrest already has. This is why there are no additional power savings methods available for Kentsfield. This isn't magic. If you cut Kentsfield's L2 cache down to match K8L's then it should match K8L's power consumption even with an onboard memory controller. Increasing cache increases power consumption.

and still be clocked high enough for single threaded stuff.

Again, you are overlooking the fact that Woodcrest has already gained by increasing IPC. This is why a 2.4Ghz Woodcrest would be as fast as a 2.8Ghz Opteron if Woodcrest were using registered DDR2 instead of FBDIMM (and if a 2.8Ghz Opteron existed). K8L is supposed to increase IPC. By how much, I'm not sure. At the moment I'm figuring 7.5% which would be about half of Conroe's current lead. This will allow an increase in performance without increasing clock. This would still put Intel one speed grade (200 Mhz) up.

Anonymous said...

I meant Intel started at 40/65/80 for dual core, more headroom to go higher when adding more cores and clock, while AMD will have to squeeze quads into their current TDP, already the max is at 120W. TDP is designated for cooling requirements, I don't see don't see what an off chip memory controller matters, unless you're talking about power consumption.

Scientia from AMDZone said...

I meant Intel started at 40/65/80 for dual core, more headroom to go higher when adding more cores and clock, while AMD will have to squeeze quads into their current TDP, already the max is at 120W.

I'll try explaining this again. Intel fits into 40/65/80 watts because it has already at 65nm and has already implemented sophisticated power control that AMD has not implemented yet on 90nm. This is not more headroom for Intel. AMD will get the same benefits when it moves to K8L.

TDP is designated for cooling requirements, I don't see don't see what an off chip memory controller matters, unless you're talking about power consumption.

Well, for servers and notebooks this adds to total power draw so you have to count it. The MC doesn't matter that much today on desktop units. However, it has been suggested that Intel will move to an on die MC in 2008 or 2009. When Intel does that their free ride on the memory controller is over.

Anonymous said...

Hey scientia isnt it in terms of performance:

***** - X6800
***** - E6700
Fx-62 - E6600
5000+ - E6400
4600+ - E6300

And not the 4200+ you place for E6300 and 4600+ you placed for E6400??

Scientia from AMDZone said...

Well, I'm confident in saying that Conroe is 15% faster at the same clock. It's possible that Conroe is faster than this but I haven't seen good enough testing to confirm this. Right now, I'd say 4200+ is a pretty good match for E6400 but E6400 looks better in SSE. I'd really like to see some good tests where the second core is loaded before the benchmarks are run on the first core. Tom's Hardware Guide did very poor multitasking testing by using a loading benchmark that was I/O intensive. It is nearly impossible to properly load the second core if the code is waiting on the harddrive.

Anonymous said...

"Releasing Conroe at low prices does put pressure on AMD but it puts even more pressure on Intel's P4 line which will still make up around 80% of the total chip volume in the last two quarters."

80% of DESKTOP! Not 80% of total chip production! As P4 is EOL Q2'07 this is a short term pressure (and shrinks overtime as P4 SKU's start phasing out at end of Q4'06).

As far as I know Conroe doesn't put uch pressure on the mobile and server chip pricing...

"This shows a fundamental difference in design style as Intel has dedicated FABs (D1D and D1C) specifically for design while AMD runs tests in its production FABs."

This is smply not true - D1c is no longer a "design" fab and simply produces "HVM" (high volume manufacturing) parts. D1d is a mix as it supports 45nm development as well as 65nm volume production (on smaller scale than other 65nm fabs).

And you might want to keep in mind AMD has only ONE 300mm fab, they have no choice but to do process development and production in the same fab! If/when AMD has a 2nd geographic site (NY), I would find it unlikely they would do development on a specific technology node at 2 different sites.

Scientia from AMDZone said...

80% of DESKTOP! Not 80% of total chip production! As P4 is EOL Q2'07 this is a short term pressure (and shrinks overtime as P4 SKU's start phasing out at end of Q4'06).

It's over 60% of server chips but these are much lower volume. C2D won't even be 30% of Intel's production by end of 2006.

This is smply not true - D1c is no longer a "design" fab and simply produces "HVM" (high volume manufacturing) parts.

As I recall, D1C hasn't been upgraded to 65nm yet.

And you might want to keep in mind AMD has only ONE 300mm fab, they have no choice but to do process development and production in the same fab! If/when AMD has a 2nd geographic site (NY), I would find it unlikely they would do development on a specific technology node at 2 different sites.

Actually, I imagine they will. The difference is not how FABs AMD has. AMD can run tests at several different FABs because of APM.

Anonymous said...

"As I recall, D1C hasn't been upgraded to 65nm yet."

Perhaps but it has been running PRODUCTION on 90nm (not design) for some time now; it hasn't done "design" for a long time. Intel has always upgraded their design fabs to production fabs after 1 or 2 generations of development.

D1a = F15 (200mm)
D1b = F20 (200mm)
D1c = "D1c" (300mm) - I don't think they renamed this
D1d is already running some production since 65nm development was completed (and will run 45nm product onces that development is complete)

Anonymous said...

I find it strange how Intel is able to run 65nm production in D1d an do process development on 45nm in the same building (in fact I've heard this crazy notion that some specific tools actual support both production and development lines).

They also managed to do this in D1c as well.

As APM is responsible for this within AMD, I'm wondering how is this possible as Intel does not have APM technology? Is it possible to do development ant production without APM? Are there other automation and process control techniques out there that may function similarly to APM?

Any thoughts?

Scientia from AMDZone said...

D1D can run production but only at a low volume. D1D is and will remain the development FAB. Actually, I'm a bit surprised that Intel doesn't have D1C upgraded inline with D1D since by being right next door it would appear that production could start almost immediately from D1D. In other words, it seems strange that this isn't the first run FAB.

No, APM does not give AMD the ability to run smaller process tests; this requires upgrading some of the tooling. Intel upgrades some tooling ahead of time to allow the smaller process tests. So far AMD has not had to do this because the smaller process testing is done at IBM. For example, AMD was delayed in 65nm die testing until FAB36 was ready; this couldn't be done on FAB30 which was only 90nm. It wasn't worthwhile to upgrade FAB30 since the 200mm tooling was obsolete.