More proof that Conroe is a one trick pony: cache
Now, we have the most direct comparison of the Core 2 and K8 microarchitecture.
DailyTech tested the Conroe-L, which is a single core Conroe with 512KB of cache. The Conroe-L 440 is rated at 2GHZ. Comparing to Athlon 64 3500+ (2.2GHZ, 512KB L2), the Conroe-L is definitely an undedog.
As you can see from the gaming benchmarks, the Conroe-L trails Athlon 64 in all tests. In one game, the Conroe-L is 13% slower. Clearly, the inherent average IPC advantage of Core 2 over K8 in desktop environment is less than 10%. I pointed out this long long time ago.
As we have seen previously, the 3GHZ Opteron frags the 3GHZ Woodcrest in both integer and floating point peformance.
K10 will simply kill Core 2.
PS: those who compare the prices of Conroe-L and Athlon 64 must note that Conroe-L is a future Intel chip no one has seen in the wild yet.
posted by Sharikou, Ph. D. @ 4/11/2007 11:07:00 PM
23 Comments:
I think this shows only one thing, that Conroe's shared L2 cache design is very effective for multi-threaded and multi-processed applications.
Oh gee what news. A lower clocked and considerably cheaper Celeron is on par with AMD CPUs. had they used comparable price AMD, 3200+, the results for AMD would have dropped by about 10%, making Conroe-L clear leader in everything besides SS2.
Talking about desktop performance,
2.0Ghz Conroe-L vs 2.2Ghz Athlon 64:
DivX MPEG2 to MPEG4: 12% faster
AVI to WMV: 3% faster
LAME WAV to MP3: tie
Audacity WAV EQ filter: 7% faster
Paint Shop JPEG noise remover: 17% faster
WinRAR file compression: 63% faster
Excel inventory convertor: 21% faster
I'd say not bad for a tweaked P3 core on a 800Mhz FSB with 10% clock disadvantage.
Sharikou,
I've been reading your blog for a few months now. I enjoyed your previous posts about sharing a positive light on AMD. However, like others have said earlier, your blog gets worse and worse in quality.
Conroe-L is a Celeron chip. If you want to prove Conroe is a crap chip then you should bring up how a Sempron would fair against this upcoming Celeron.
"Conroe-L is a Celeron chip. If you want to prove Conroe is a crap chip then you should bring up how a Sempron would fair against this upcoming Celeron."
Celeron (CONROE-L) is still native conroe chips, only the L2 cache that makes it different with the others conroes. Please correct me if im wrong.
You are perfectly correct but the fact remains that it is supposed to replace Netburst celerons.
Semprons are no different from other K8's too, only they have less L2 cache. I wonder why weren't they used in the comparison.
HoHo
I wonder why weren't they used in the comparison.
Here is a link to an HKEPC article showing the processor against Athlon and Sempron.
Sharikou
Why is it you don't show an article like this one which compares an Intel Core 2 Duo E6850 and a AMD Athlon 64 6000+?
Oh, thats right it would be a fair comparison, not a Celeron to a Athlon.
You bunch of freaking CRY BABIES!!!
Name it Celeron, Athlon or whatever you like, it WON'T change the fact that this is a similar CPU to CPU comparison which features similar price, similar cache and similar freq.
Sure the Athlon has 10% freq. advantage and HT, but we can also argue that this a 4 year old architecture against a brand new one too!
Bottom line is this is a "similar priced" processor competing for the "same market"... period!
If they compared a $30 AMD processor to a $60 Intel processor, would that have made you happy?
jeach!
"this is a similar CPU to CPU comparison which features similar price, similar cache and similar freq"
If you allow "similar" to have rather big differences then you are correct.
Funny thing is that C2 has 2M of L2 per core. That CL has 4x less. x2 have 512k per core, 1/4 of that is 128k, the same amount that K8 has for instructions and data L1 cache combined. I'm not trying to say anything, just that this CL has massively crippled cache and AMD's don't.
"Sure the Athlon has 10% freq. advantage and HT, but we can also argue that this a 4 year old architecture against a brand new one too!"
Some might say it is K8 against beefed up ancient P3. Also wasn't K8 supposed to be far superior to anything Intel would ever have, or at least would have before its BK 2008?
"Bottom line is this is a "similar priced" processor competing for the "same market"... period!"
I guess you could say so, allthough there will be Pentium E series with 1M cache that will replace current singlecore netburst Pentiums. Against what would they fight?
Bottom line is this is a "similar priced" processor competing for the "same market"... period!
Since the 2.0Ghz Conroe-L will be introduced for $59, it should be compared to the $58 2.0Ghz A64 3200+, rather than the $69 2.2Ghz 3500+.
Of course, the $73 X2 3600+ would probably be a much more sensible choice than either...
This comparison makes sence only if beefing the cache is a trick AMD is keeping as a winning card. If this is true, then all athlons X2 512K chips become AMD chip one step above current semprons, and this comparison in this blog start to make sence.
In the other hand, it could be that bigger cahce does not make a big difference in performace in AMD CPUs, making this benchmark just a bogus.
Only time will tell.
al-saqer
"This comparison makes sence only if beefing the cache is a trick AMD is keeping as a winning card"
Let me remind you that there is almost non-existent speed difference between 1M and 0.5M L2 AMD CPUs. Even 0.25M ones weren't that much slower. This is thanks to IMC and massively better memory latency, a thing Intel doesn't yet have.
Now who wants to guess how much will Intel CPU performance increase once it gets IMC in a couple of years?
Jeachass said; Bottom line is this is a "similar priced" processor competing for the "same market"... period!
When AMD outperforms anything intel has for the SAME PRICE doesn't that reflect poorly on AMD?
That just means that AMD has to do/offer/spend more to get a sale.
Intel on the otherhand sells everything with some level of premium.
2.40 ghz Conroe-L vs 2.20 ghz Athlon64.Conroe-L is equal performance with singlecore Athlon64.I see larger cache is rising performance a lot.
When AMD outperforms anything intel has for the SAME PRICE doesn't that reflect poorly on AMD?
Yes it does, so long as that same price is targeted for the same market.
What I mean by same market is if Intel offered Itanium processors for $70, you couldn't compare it to the 'value' market because it would cost way too much to get a whole system regardless of the processor discount.
There are no written rules, just common sense... of course!
That just means that AMD has to do/offer/spend more to get a sale.
Intel on the otherhand sells everything with some level of premium.
Why do you think AMD is currently loosing money and Intel is still surviving?
Althoug, all of that is changing... quarter by quarter... AMD is growing!
"When AMD outperforms anything intel has for the SAME PRICE doesn't that reflect poorly on AMD?"
I think that reflects more poorly the intelligence of those who buy Intel processors. :-)
Althoug, all of that is changing... quarter by quarter... AMD is growing!
Losing market share, lower revenues, lower stock price and zero profit.
How exactly is that growing?
Simply as noted: Celeron costs 50 bucks and compared against a similar to slightly more expensive AMD offering it is close. Priced to value, priced to comptetion, priced to performance, priced to BK AMD.. haaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa!
You compare a Camry to a Accord and an Lexus to a Acura. So what if the Camry and ES350 are the same platform. The bottom line in the high end is still performance/buck/power.
You got a Phd or are you a "Ph"ony "D"octorate. You sure sound like a Phony to me.
On a seperate note:
Listen carefully what is said and what is NOT
AMD's European technical director, said: "Barcelona is an architecture that must be faster."
Why does he say "must" versus "is" It must be faster because if it isn't they AMD is finished. While if he says it "is" that means he has seen the data and based on real measurements "is." Must is hope, required, management FUD of what is desired vs. what reality is. Its like Hector saying we "must make money" versus we "are" making money. Notice he can't use that verb. Listen to the verb used fanbois.
He laters goes on record saying.. " much depends on applications." Does that mean he has seen the benchmarks and that Barcelona doesn't win ALL benchmarks. How can something soo good with better cache and faster cache, 4 cores on a single piece of silicon, with IMC not be better then INTEL Core2 across teh board.
INTEL and AMD numbers will be out next quarter. AMD is way down and will suck red. INTEL will be down a bit YoY as the price war take a piece of revenue and profit. The biggest joke on the AMD fanbois will be that INTEL will still make hundreds of millions of dollars while according to Sharikou selling CPUs at fire sale prices.
There's no denying AMD is the architecture King and Intel is a hack job.
For the last three to four years AMD has dominated in performance and price. See the url below for a classic example.
http://www.anandtech.com/cpuchipsets/showdoc.html?i=1927
Low end Duron 1.6GHz outperforms a much higher clocked and 2X more expensive Celeron 2.6GHz. It also outperforms a Pentium 4 1.8GHz at over 3X the cost. The domination continues with Athlon XP, 64 and X2. Only recently did Intel's new Core2 architecture with much larger L2 cache caught up to AMD's two year old technology. AMD will soon demonstrate again with its true quad-core Barcelona that superior architecture is king without relying on bloating the cache to make up for deficiences, upping the clock significantly or other hack jobs.
AMD's financial situation is very dire. Ati and AMD together are going to report less revenue than AMD alone reported this time last year. AMD's marketshare is crumbling. No one wants AMD's antiquated processors when they can have Core 2 Duos. AMD is forced to sell it's 3600+ processor at $65, they're not even making a profit at that price. Meanwhile Intel continues to make profit, even while pumping billions into it's 45nm technology.
Barcelona is too little too late. The current 3Ghz Clovertown will easily frag Barcelona at it's measly 2.3Ghz introduction speed.
Later this year Intel will introduce Penryn, an updated architecture that will frag AMD even more. Penryn will just widen the gap even further. AMD will never be able to catch up. AMD will still be selling 90nm processors when Intel introduces 45nm CPUs. Going 45nm will allow Intel to cut prices even further, forcing AMD further into the red and leading to their BK. Intel won't even need Nehalem to finish AMD off.
AMB BK Q2'08.
How exactly is that growing?
Growth isn't necessarily "hard assets" (cash, FAB, chips, etc).
And certainly not represented in share price.
A little while back I wrote about the cycle in the operations of a company!
Starts with CASH > R&D > PRODUCT > SALES > ACCOUNTS RECEIVABLES > and back to MORE CASH
Many companies have a hard time with this cycle. The faster you can convert from one to the other, the better.
For example, look at DELL, they have tons of money stuck at the ACCOUNTS RECIEVALBE cycle.
Anyway, my point is that AMD has many patents and products in development right now which has 'burned' the cash, but is not yet in sales. They have many, many products stuck in these cycles. Once they start moving to other cycles, you'll see how much cash they can generate.
Another form of 'growth' is the addition of capacity. AMD has paid for the FAB's and tooling but because AMD does not utilize all of this capacity yet (full 65nm ramp), the potential is there but not being used. Once again, its just a matter of time.
So due to the above two examples, AMD has may have a small market share retraction, but only because of having all its eggs in the same cycle. Once things move along, you'll see it.
So for AMD to look good you have to compare their top range series to Intel's bottom range series,
K8 vs Celeron?
Post a Comment
<< Home