Are these benchmarks for SPecFP2006, or SpecFP_rate2006?
There is a huge difference.. If it's based on the former, then it could be a pretty good indication of how the K10 will perform on the desktop.
Are these benchmarks for SPecFP2006, or SpecFP_rate2006?
There is a huge difference.. If it's based on the former, then it could be a pretty good indication of how the K10 will perform on the desktop.
How old were you back then if you have such a poor memory ?Originally Posted by Haltech
K7 and P3 were on par IPC wise ( gains were ussualy were SSE/3DNOW was being used ).Late on P3's life , K7 managed to out do it , MHz wise.
Early P4 ( Willy core ) sucked = > early 2000 to early 2002 AMD on top. Anyway , their succes was shadowed by the crappy chipsets they had.
Only in 2004 K8 managed to out do the top Northwoods and the victory was obvious with the advent of dual cores in 2005.My oh my how things turned in 2006Circa A64.. Same thing repeats.
So out of 25 years of x86 history we have 4 years with AMD on top.Go eat your bananas now.
Ed's article is worth a look :
http://www.overclockers.com/tips01142/
This post on another forum is right:
http://www.siliconinvestor.com/readm...msgid=23484492
>>Today's AMD PR claims 50% in floating point and 20% in integer at same clock. So even if there is 3ghz (Intel) vs 2.5 ghz (amd) frequency advantage, that would still translate into a 25% floating point advantage and a tie for integer.
Oh, no. You see, it is "up to", meaning the very best one they can find, in both cases. Furthermore, it is clearly the "rate" versions of SPEC_int/fp 2006 in both cases, in a 2P configuration.
AMD will lose virtually everything against 3GHz Woodcrest, outside of bandwidth-intensive FP apps, and that is once they manage the 2.5GHz parts. And don't forget they are stuck at 2.3GHz 95W at launch for a quarter, if DailyTech is right...
Penryn @ 3.33GHz (or higher) is going to blow Barc out of the water.
--------------
You want another indication for how bad this is for AMD? Look at K8 vs. same-clocked C2D on 2P rate for 2006 fp/int:
int: 3% faster
fp: 16% faster
And yet, consider the huge variety of applications where same-clocked Woodcrest/Conroe *blows away* same-clocked K8, despite the 2006 Spec int/fp rate marks!
So, how does a Barcelona core compare to a K8 core clock/clock?
"Up to" 17% faster for int
"Up to" 29% faster for fp
Trouble is: 95W K8 is 2.8/2.3 : 22% higher clocked than Barc.
120W K8 is 20% higher clocked than Barc.
So Barcelona integer performance *will not improve* from K8 levels, once clock is taken into account.
FP performance gets a bump, as expected.
Conclusion: 3GHz Woodcrest will continue to dominate on all apps except bandwidth-heavy FP stuff.
Then Penryn arrives.
------------------
I would add, it goes without saying that Woodcrest will maintain a large lead in gaming.
The Athlon 64 was launched in September of 2003, not 2004. Before that, blows were roughly traded back and forth between intel and AMD.
You're also forgetting the many times intel made an attempt to block AMD from competing completely. 1986/87, 1990, 1997, and 2005 all had lawsuits filed. The 1990 lawsuit being one filed by intel, yet won by AMD being one of the more important ones, because it was an attempt to block AMD from 486's. The 1986/87 was important as well, because intel attempted to stop amd from producing chips(which was a breach of contract btw)... So, out of 25 years, how many of them were strong armed by intel? As you know, a lawsuit doesn't occur over night, it takes years to get up the evidence to sue in the first place.
Sounds like a few people may have their history a bit messed up.
Interesting article..
Is AMD's supposed lead in FP based on vector performance?
It would have to be, because the regular x87 FPU remained untouched as far as I know.
Although C2D has a higher theoretical throughput than the K10, I guess the k10 will be much more efficient in that regard.
One thing is clear though.. AMD needs to migrate to 45nm as quickly as possible, and when they do, they need to make sure that the 45nm version of the K10 is more than just a shrink.
Intel's Nehalem will undoubtedly have awesome performance, and will most likely outperform the K10 unless they make substantial improvements when they shrink it to 45nm.
^^ What |JACK- said.
Really ? And how did the 1.8GHz K8 fare against a 3.2GHz Northwood ?
Not well if you remember.
Only in 2004 did the K8 managed to get in front...
Intel blocked AMD from reverse engineering its chips , which is perfectly legal.You're also forgetting the many times intel made an attempt to block AMD from competing completely. 1986/87, 1990, 1997, and 2005 all had lawsuits filed. The 1990 lawsuit being one filed by intel, yet won by AMD being one of the more important ones, because it was an attempt to block AMD from 486's. The 1986/87 was important as well, because intel attempted to stop amd from producing chips(which was a breach of contract btw)... So, out of 25 years, how many of them were strong armed by intel? As you know, a lawsuit doesn't occur over night, it takes years to get up the evidence to sue in the first place.
Sounds like a few people may have their history a bit messed up.
Sounds like some have a weakness for the underdog whick blocks their viewing angle...
Like what ?
If you remember AMD played all its cards , Intel still has 2 aces : IMC and CSI. What if Intel gains 40-50% only due to IMC+CSI ?
What can AMD do to match that ?
On par my ass. K7 cleaned the clock of a P3 without any sweat whatsoever, clock for clock, not to mention, the FIRST to hit the 1 GHZ barrier. Intel tried to pull a 133 fsb upset, but it didnt work. K7 NEVER lost the performance crown to a P3,P3 Coppermine NOR the Williamette version of the P4... Not even from the beginning. The K7 was intels wakeup call to AMds previous failed attempted to gain the stronghold. The chipsets werent the problem. it was the MB manf hiding the fact they were making boards from Intel... Asus whitebox anyone? K7 tested the current power supply limits, which had an affect on the stability of a machine. Some others had used cheap generic DDR modules that caused compatibility issues.
Your intel fanboism shows greatly and that is truely sad.. I can recall in the EARLY 90's where the AMD DX 100s were slapping Intel around like a step child.. I also find it hard to believe that Intel has had 25 yrs of success as the fasted chip on the market.. Try the Motorola 68XXX series chips that ate Intel for breakfast. Commodore Amiga, Atari ST and Macs all used these processors. Intel sat in the backseat...Thats how old i am...
Well, I'm no engineer so I can't say what AMD needs to do..
But, they have a few years head start on Intel when it comes to IMC and direct connect bus..
There has to be other avenues left to explore though, and once AMD gets on 45nm, they'll have more space on die to add more features to the processor.
AMD didn't launch a 1.8 ghz A64 first...
At launch was the 3200+, the FX-51. The 3200+ was a clawhammer at 2 ghz, and fx-51 was at 2.2 ghz.
@carfax
the benchmarks are specFP_rate2006 and specINT_rate2006 on a 2P servers.
Because of the ODMC and the point-to-point buses, the 2P K8 performance scales much better than 2P C2D.
The dualcore K8 is 2.5% faster than C2D in specINT_rate2006, at same frequency, while it is 15.5% faster in specFP_rate2006.
If we consider AMD's bold claims as true, then the 20% advantage over C2D in specINT_rate2006 will become 17% advantage over K8, while the 50% in specFP_rate2006 will become 29% advantage over K8. In average the K10 is 23% faster than K8.
But, these benchmarks are on 2P systems, where K8 shines and scales much better in performance than C2D, because it has an ODMC and a point-to-point buses.
On the desktop, the K8 has no advantage over C2D(no core-FSB-northbridge-FSB-core communication) by having a point-to-point bus. The 1P K8 system has no additional bandwidth provided by NUMA.
The average C2D performance advantage is 20%, compared to a same clocked K8. According to AMD's specXXX_rate2006 estimations, clock for clock, K10 will be 2.5% faster than C2D in average.
@savantu
Stop dreaming. IMC + CSI can't improve performance for 40-50%
For example, K7(Barton with FSB400) vs K8. K8 is a tweaked K7 core with ODMC and new point to point bus. It is only 10% faster clock for clock, due to architectural improvements.
Last edited by gOJDO; 04-23-2007 at 10:17 AM.
And not only are they for a 2P system, but that is *at the same clock*. But the K8 runs at 3GHz to K10's 2.5GHz, and 2.8GHz to K10's 2.3GHz (for 120W and 95W TDP bins)
So really, no int_rate improvement at all for K10 vs. K8, and a small FP improvement for K10 vs. K8, *because* the clock rate has fallen as IPC has improved. Now, of course, this is all per core, and K10 brings quad core. But it's going against Intel QC at 3GHz Woodcrest and 3.33GHz+ (Penryn).
If only AMD would release actual scores, with actual clocks and system configs. Why can't they do this?
I own both, so a fanboy I'm not. Why am I saying that AMD is full ofUmm Circa 1999.. Athlon blows Pentium 3 out of the water...
Circa A64.. Same thing repeats.
Who are YOU to say who is full of and who isnt? Wait for the release. Either you will eat a shovel full of your own fanboy , or you walk away with no turds on your bib. Either way, being a fanboy is the lamest thing you can be.,? Who's seen these benches beyond AMD walls?
Release some god damn benches by now and throw us a freaking bone.
AMD doesn't seem to understand their userbase. Sure us hardcore freaks account for maybe 1% - 2% of their total sales, but that 1% probably represents 10% or more based on word of mouth. I build atleast a hundred systems a year. They used to be about 85% based on Athlon 64's; now that number has dwindled to about 15%. I'm one person out of thousands who now recommends Intel systems. If you want me to tell my customers/friends to wait for something better from AMD, then let me know what's coming down the pipeline and what it's capable of.
I don't need to know every thing about the chip, but some nice video/audio encoding benches mixed with a few game benches would do just fine.
Last edited by freeloader; 04-23-2007 at 10:52 AM.
As quoted by LowRun......"So, we are one week past AMD's worst case scenario for BD's availability but they don't feel like communicating about the delay, I suppose AMD must be removed from the reliable sources list for AMD's products launch dates"
^^ Ye, the HT/CSI is 100% useless on the desktop. And really only benefitting on 4S+, maybe even 8S+ first with the new quad FSB platform.
IMC is a semi joke. Core/Core2 already showed an IMC aint a winner. Also for those doing their homework, would know things like 1066->1333FSB gives 1-3%. There simply aint much to get there.
However, the IMC is a must for the HT/CSI style. But else its just a little useless checkbox for the desktop.
Ofcourse I wish as many others a system on a chip. Simply due to simplicity of it. But dont expect/hope it gives anything but cost savings on the desktop.
Crunching for Comrades and the Common good of the People.
shame on youcurrently amd has the best performance/cost ratio - and you recommend people who obviously wil not OC expensive intel-based systems
people noticed windows run smoothly on imc-based cpu
AMD isnt like Intel. AMD wont release leaks of benches until a week before its launched. Besides, we make up and extremely 1% of the market, so AMD could care less about what we all think on these OC forums.I own both, so a fanboy I'm not. Why am I saying that AMD is full of
Release some god damn benches by now and throw us a freaking bone.
AMD doesn't seem to understand their userbase. Sure us hardcore freaks account for maybe 1% - 2% of their total sales, but that 1% probably represents 10% or more based on word of mouth. I build atleast a hundred systems a year. They used to be about 85% based on Athlon 64's; now that number has dwindled to about 15%. I'm one person out of thousands who now recommends Intel systems. If you want me to tell my customers/friends to wait for something better from AMD, then let me know what's coming down the pipeline and what it's capable of.
I don't need to know every thing about the chip, but some nice video/audio encoding benches mixed with a few game benches would do just fine.
If you notice, Intel has a much more influenctial crowd than AMD. A lot of guys here have inside connections, where AMd is more locked down.
Gimme a break. I tried both. There is no difference. Its nothing more than a fud. And I never heard or seen it anywhere else.people noticed windows run smoothly on imc-based cpu
Maybe you should send it to fudzilla or theinq.
Crunching for Comrades and the Common good of the People.
Only for the low end...shame on you
Crunching for Comrades and the Common good of the People.
i've heard at least one K10 sample has been beeing to Moscow, RUS for some time
This is not true!shame on you
Some AMD CPUs have better price/performance than some Intel CPUs and vice versa.
http://www.xbitlabs.com/articles/cpu...e-roundup.html
WOW! Really, there are a lot of very stupid people.people noticed windows run smoothly on imc-based cpu
E6600 and 6000+ seem not to be low end
PS new intel prices are not anywhere now
for instance now i can buy E6600 for 320USD, 6000+ for 250 USD
Last edited by MAS; 04-23-2007 at 11:09 AM.
Who cares about stock clock
...
Posting Permissions
Reply With Quote
Bookmarks