Printable View
http://www.tgdaily.com/content/view/30928/135/
Compared to Intel's fastest Xeon 5300 (X5355), AMD promises that will be about 40% faster in some applications (best vs best).
http://www.tgdaily.com/content/view/31729/135/
20% in integer, 50% in floating point (at the same clock). That 50% could come in handy if you do a lot of POV-ray :rolleyes:
Some HPC crowd would strongly disagree with you red lol.But,it seems that intel fanboys will push the "not fast enough" mantra once the K10 hits the shelves.It will never be fast enough ,since it's not made @ intel (sarcasm)
GoThr3k said that 40% is overall. Maybe for supercomputers, but not for the other 99.9% :rolleyes:
http://www.spec.org/cpu2006/results/...403-00786.html
http://www.spec.org/cpu2006/results/...430-00981.html
http://www.spec.org/cpu2006/results/...412-00810.html
http://www.spec.org/cpu2006/results/...528-01175.html
As core count increases, Intel loses its lead in the good ol' spec benches at 8 cores and you can see that in specfprate, AMD's best result is already 62% ahead of Intel and 2% on int. So they are extending an already big gap not making a new one.
hold it there intel Marketing dudeQuote:
Originally Posted by red
I said that AMD said that....
And the links you post are just pathetic...
twin brother of shintai?
Again, AMD said 40% in some, not overall. And 20% in INT, 50% FP, and in real world usage, INT dominates. How do you get that AMD said 40% overall?
they said it somewhere, a couple of months agoQuote:
Originally Posted by red
but 20% int, 50% FP is also a thing they've said
Clearly they never said 40% overall, unless you have really optimistic interpretations of "40% in a variety of ??workloads", "performance per watt", "some apps".... BTW, the 20% number is outdated. http://multicore.amd.com/us-en/AMD-M...rformance.aspx It is comparing results as of April 16. The best Intel specintrate result as of then was http://www.spec.org/cpu2006/results/...219-00526.html and since then they have gotten http://www.spec.org/cpu2006/results/...528-01177.html The 2.6Ghz K10 is 2% faster than the 2.66 Clovertown (102 vs 99.9).
where do you get the 2.6 k10 results from?
Umm, the chart from amd.com? With the 1.2x, 1.5x figures from AMD, multiplying those factors by the best 8 core Intel results in spec int/fp rate before 4.16 gives basically what the graph suggests, 84.8*1.2=101.7, which is less than the max of the graph of 104, and http://www.spec.org/cpu2006/results/...205-00350.html
58.8*1.5=88.2, less than 92. BTW, 101.7 is 1.8% better than 99.9 lol
Red there are no official results for K10 AND we do not know for sure how it will perform.We have vague estimates from AMD.
K10 is way ahead of K8 in arch. improvements and it will be faster than 2 core and 4 core intel based systems(clock for clock ).You will see that when it gets reviewed, i can tell you that much.And Penryn won't change that much.Nehalem probably will,but than again,k10 has an evolutionary path also(shrink) and a 8 native core part in H2 2009 is coming from AMD.
I don't think I said they were official. It is all guesses, as is this whole thread. But as you have said, compiler, configurations, etc. can greatly swing the results. It is unknown whether the scores provided are on the hopeful side or sandbagging.
And you have the nerve to call out my analysis :rolleyes: If all you have is hype and no numbers, I could frankly care less. BTW, gl on AMD with their 1.6GHz dual/quads beating Intel's 1.6GHz dual/quads lol.Quote:
K10 is way ahead of K8 in arch. improvements and it will be faster than 2 core and 4 core intel based systems(clock for clock ).You will see that when it gets reviewed, i can tell you that much.
maybe its a bold statement, but if you look at the improvements, it should be better imo IPC <-> IPCQuote:
Originally Posted by red
But we will have to wait for another few months to see the results, let's hope they're good so that AMD can breathe again
Actually, the new, "demanding" apps are running SSE code, not FP code.Quote:
Originally Posted by LOE
And don't tell me they are the same, because they are very different. FP code has been left in to essentially support old applications that rely on it. It is notoriously hard to optimize in assembly language because the instructions for it have variable lengths.
This is one of the main reasons SSE/2/3/4 were developed. They have fixed length and can perform the same mathematical calculations with instructions that are all the same size. This is why it is so much easier to optimize SSE code for speed at the assembly level - all the instructions are the same length.
And since there already is a very large installed base of SSE code, this is where K10 is going to see the biggest improvement over K8.
Here's what Johan from AnandTEch has to say about AMD's K10 perf. preview's that were reported in some "fast news" websites:
http://www.aceshardware.com/forums/r...3522&forumid=1
So K10 is not even near those levels in retail version,it's much much better.Quote:
Originally Posted by Johan@AnandTech
What a useless thread it became :(
K10 is the same speed as K8 for scalar SSE. It is only vectorized SSE that is faster.Quote:
Originally Posted by LOE
No, it got the same SSE units, just twice as wide.Quote:
Originally Posted by LOE
http://www.spec.org/cpu2006/CINT2006/
http://www.spec.org/cpu2006/CFP2006/
I frankly don't know what "int" or "fp" is, except that I can tell that according to the int and fp that AMD is talking too, the only thing remotely real world on the fp list is ray tracing and speech recognition:rolleyes:
An as informal has just shown, AT and DT are seperate entities with different think tanks. So where is the AT hate coming from? BTW, I rather like the sloppy and dirty benches DT bring:D And I would hope that performance, and we can be hopeful with some s7 guy saying 10% better same clock, metro.cl and cooper saying yay, http://www.theregister.co.uk/2007/03...ona/page2.html The Register hearing sources beating their chest, but recent evidence suggests it is hampered with bugs, delayed, not clocking high enough, soon to be fighting an early Penryn not Conroe, etc.
http://www.softlayer.com/servers_dpmc_compare.html
http://www.softlayer.com/downloads/SL_Opteron2000.pdf
http://www.softlayer.com/servers_dpmc.html
Opteron 2000 QC Series Available July 07
Speeds listed from 1.9-2.4.
Strange, 8 x 512k for 8 cores (dual socket system).
They just didn't list the L3 cache..It's there.
Barcelona is coming :DQuote:
Originally Posted by red
AMD: Barcelona still on track for a Q3 announcement
http://www.tgdaily.com/content/view/32533/135/