sailor said:
Conclusion : The K10 archtecture/memory control offers serious advantages over Intel when it comes to Running multiple, heavily multi-threaded, processor-intensive applications simultaneously - like BOINC !

Let's see how K10 does on benchmarks that people were able to check:

On Povray: (perfectly threaded) 1st Benchmark choosen by AMD, probably one of their best case!!!!

obviously faster ...


A serious advantage, sure ...

nothing better than a little video on yourtube

on cinebench:: (perfectly threaded)
very very fast

but not enough it seams.

on heavily multithreaded application, K10 gets smoked! see data in the articles :)

and it is already 1 millions of them out there!

those are the only available benchmarks publically up to TODAY! the rest is AMD powerpoint presentation and their "projection in lalala land", face it!

Sailor, that's enough solid data for you?

who?

Those are Very recent links, can't be more accurate!

who?

cant u read ? :S

SPECint_rate2006 - 21% AMD there is ure INT btw
SPECfp_rate2006 - 50% AMD
SPECweb99 - 67% AMD
SPECweb99_SSL - 66% AMD
SPECweb2005/Ecom - 100% AMD
SAP-SD - 70% AMD
TPC-C SQL2000 42% AMD
"lalala land projection, nobody did measure this, not even AMD, it is PROJECTED ;-) , do you know anybody who will project losing????"
stop blowing hot air, when numbers are already called.

When you feel you have some numbers to add, then do so, and dont write "Strange they forget to speak about those ..." cuz this tells nothing at all...

PS: funny that u didnt comment on my facts about realibility - truth hurts ? :)

Next month i might put an AMD x2 6000+ up, in benches it beats the C2D 6600 and its listed with 139 € - can anyone beat that atm ? :o
DC wont be the main use of that system, but ill see what RAC I can pull :)

those are projection numbers, they actually did not file them in SPEC database. no more comments!

who?

You may beat a 6600, but you will not beat any of the X6800 or Q6600 for sure, thanks for admitting this. I don t even speak about Penryn ... no chance to even get close here! yes, it is already played

smile

who?

now, journalist agree with me. When the Journalist is actually a sharp one, here is what is coming out: George Blog

In the Pentium 4 time frame, George was hard on Intel too. He is a fair man.

who?

This random argument has got over 600 views :O

---

This random argument has got over 600 views :O

You've never heard of orderly chaos?

This random argument has got over 600 views :O

You've never heard of orderly chaos?

it's like a bunch of sharks circling a chunk of chum.

---

Dailytech has its word to stay about how Naive we are ...
DailyTech

simulated ... hahahhaa


who?

Hm, there are some things that dont fit in my point of view.
On one hand you talk about simulated-non real benchmarks, but when we go to the POVRAY score, where Intel claims to beat 16 core AMD Barcelona with a 8 core Xeon System, you take those numbers as hard evidence ;) ?

The cinebench was run on a 1.6 GHz Barcelona comparing to a Xeon X3220 , 2.4 GHz. And also giving the Xeon faster memory. The Barcelona can make more use of faster memory, like i quoted earlier, cuz of their advantaged memory controller. The result can be compared better this way I think : the Kentsfield Xeon was 58% faster with a 50% higher clock frequency - and we all know, rising the clock rises the performance progressivly, not linear. So in no way I see a win for Intel in that benchmark.

And if the SPEC numbers for the Xeon are really old/outdated, and the actuall one is 101 for the 2,66 GHz, then the 2,6 GHz Barcelona still hold the 102 - doesnt look that good, but 1 point is 1 point, no ?

What also will bring a performance boost for AMD will be the introducing of HT (HyperTransport) 3.0 in 2008, which will raise the data speed up to 41,6 GB/s - how fast is Intel´s FSB, even with 1333 MHz ? anyone ?

You can turn it like you want, the AMD Barcelona IS a hard opponent, and offers advantages over Intel in many sectors; ( f.e. Memory Bandwidth, the Memory Controller) - if it would be such a bad CPU like you claim, why would you go and waste your time trying to talk it down ? :)

---

http://www.MIAteam.eu

I don't see the point to be honest of throwing old or speculative articles at each other. as long as nothing is released and testable by the public, I won't believe any of the articles, may they be from AMD or Intel. Journalists are bias too, and as long as we don't see a competition of those new chips here on BOINC, we won't know what they can and cannot achieve.
Let's wait for the first Barcelona chips appearing here, cause until then it's all marketing: AMD wants people to wait for their new product and Intel wants to sell as much as possible before.

---

Hm, there are some things that dont fit in my point of view.
On one hand you talk about simulated-non real benchmarks, but when we go to the POVRAY score, where Intel claims to beat 16 core AMD Barcelona with a 8 core Xeon System, you take those numbers as hard evidence ;) ?

since I am the one who ran the numbers on the V8, and the score of 4000 for Barcelona was given by an AMD VP (Randy Allen ), I think we can trust Both numbers. May be you want to challenge Randy? hehehe, for my Povray score, more than 20 journalist saw it. So, yes, the V8 did give a serious lesson to the 16 Cores 1.8Ghz Barcelona, it is a fact you can t argue, there are more than 20 journalist able to confirm, they were at Both demos at Microprocessor forum.

And if the SPEC numbers for the Xeon are really old/outdated, and the actuall one is 101 for the 2,66 GHz, then the 2,6 GHz Barcelona still hold the 102 - doesnt look that good, but 1 point is 1 point, no ?

This is still very far from the 21% you was talking about, and the 3.0Ghz score for clowertown is filed. and the 2.6Ghz barcelona does not exist.

What also will bring a performance boost for AMD will be the introducing of HT (HyperTransport) 3.0 in 2008, which will raise the data speed up to 41,6 GB/s - how fast is Intel´s FSB, even with 1333 MHz ? anyone ?

By the Time AMD get the HT 3.0, Penryn is selling, i don t even speak about Nehalem, coming with CSI that is much smarter then HyperTransport 3.0.

You can turn it like you want, the AMD Barcelona IS a hard opponent edit(Hard to make, so hard they can go over 2.0Ghz), and offers advantages over Intel in many sectors; ( f.e. Memory Bandwidth, the Memory Controller) - if it would be such a bad CPU like you claim, why would you go and waste your time trying to talk it down ? :)

well, if you are an inform programmer, you ll remember that in 1 year of computer science class, in the architecture part, your teacher told you that you do NUMA when you can do anything else.


Please show a real benchmark, not projected, not simulated, or stop FUDing, only speak with fact please.

who?

Even AMD agree with me !!!! wow!

George said: "A day after I blasted AMD for posting deceptive on Barcelona, AMD is promising to release “honest Barcelona benchmarks“. As of Thursday afternoon 7/5/2007 the misleading Barcelona numbers are still posted on AMD’s website but AMD is promising to release updated numbers in the coming weeks with Barcelona 2.0 GHz scores (slated for September 2007 release) instead of “simulated Barcelona 2.6 GHz scores”. "

wow! here too,VNUNET

who?

touche'

Even AMD agree with me !!!! wow!


who?

touche'

Even AMD agree with me !!!! wow!


who?

The Story is not finish: AMD drags feet on taking down bogus benchmarks

really! who can support this kind of behavior?

Naive Quad core, with not so Naive Marketing, big lies! I wonder when the people who bought stocks based on those Benchmarks will start calling their lawers?

who?

touche'

Even AMD agree with me !!!! wow!


who?

The Story is not finish: AMD drags feet on taking down bogus benchmarks

really! who can support this kind of behavior?

Naive Quad core, with not so Naive Marketing, big lies! I wonder when the people who bought stocks based on those Benchmarks will start calling their lawers?

who?

Finally, AMD heared the voice of the reason:
George posted: Here
[UPDATE - 1:00 PM - AMD has done the right thing and removed the objectionable slides this morning. Thank you AMD. Blog title was changed from “AMD drags feet on taking down bogus benchmarks”.]

Fair competition is good!

who?

Who?, the POVray benchmarks. Where they using code you optimised for the Core2 or generic code...

Did anyone compare it to an AMD hand optimised code ?

---

Team mauisun.org

Maybe we should all stop posting, or even reading, Who's gloating and ego stroking threads. I'm sure he would eventually stop when he see's no one really cares, or even looks at these "Intel is god" threads.

---

Maybe we should all stop posting, or even reading, Who's gloating and ego stroking threads. I'm sure he would eventually stop when he see's no one really cares, or even looks at these "Intel is god" threads.

You're like the guy who "Replies to All", saying "Don't Reply to All!"

=;^)

---

Reno, NV
Team: SETI.USA

Maybe we should all stop posting, or even reading, Who's gloating and ego stroking threads.

...a matter of perception and interpretation of the language.

I'm sure he would eventually stop when he see's no one really cares...

I care when points of view are presented in an attempt to illuminate or educate... let the readers sort out whether there is any validity in what is presented.

Who?, the POVray benchmarks. Where they using code you optimised for the Core2 or generic code...

Did anyone compare it to an AMD hand optimised code ?

It actually make a very little difference. The SSE2 version of Povray uses Scalar SSE2, it dose behave very close to x87, only 2% of the SSE2 FP instruction retired are Packed.
If you compare the both speed on a Core 2, you 'll see a 5% difference.

The K10 and the K8 share the same number of execution units, the only performance improvement on the K10 will come from Packed instructions.
Povray does not have too many of those, and it is not FP load banded too. So, K10 = K8, it is expected on Game , X87, and SSE2 scalar.
3 FP units 64bits = 3 FP Units 128bits if you have only Scalar or x87, that is pure architecture analysis, from the public documents of K10, that is the only think I can say.
K10 has an additional FP load port.
AMD people are smart, i don't think they will choose the slower version, as 1st public demo, I wonder why people under-estimate them on this.

Got the point?


Who?

Maybe we should all stop posting, or even reading, Who's gloating and ego stroking threads. I'm sure he would eventually stop when he see's no one really cares, or even looks at these "Intel is god" threads.

Strong opinion ... hehehe please put as much energy in getting your RAC over 100 :)
The right to have strong opinion comes with higher RAC.
At least, i work hard in making my ideas into silicon, what about you? anything useful to say or do?
who?

Maybe we should all stop posting, or even reading, Who's gloating and ego stroking threads. I'm sure he would eventually stop when he see's no one really cares, or even looks at these "Intel is god" threads.

Strong opinion ... hehehe please put as much energy in getting your RAC over 100 :)
The right to have strong opinion comes with higher RAC.
At least, i work hard in making my ideas into silicon, what about you? anything useful to say or do?
who?

Easy big guy, your sounding a bit frustrated and your wording is sounding almost like a personal attack on the poor soul who is expressing more or less the same opinion I have now about this thread. Lots of chest thumping and little rational talk.

---