First solid Barcelona numbers tip up

[informal]

Surfaces from an alien spot

By Charlie Demerjian: Wednesday 28 February 2007, 11:58
IT LOOKS LIKE the first Barcelona number have surfaced at Seti@Home of all places. If you look at BoincStats, you will see what I mean.

Look at the number one CPU, an 'unknown' AMD CPU that has no listed clock but two cores. On the last column, average credit per CPU, it kind of kills the Intel CPU in the number two spot on performance.

Please bear in mind that this is just one app, is based on a semi-anonymous hackable format, and can be skewed in all sorts of ways. That said, it is an interesting data point.

Thanks to Dresdenboy at Aces for the link. µ

http://www.theinquirer.net/default.aspx?article=37907

Maybe it's K10 dual core variant,maybe it isn't.But if it is,it may be working at low clocks(ES) and producing awesome results^^.

Here are the results:
http://www.boincstats.com/stats/host...sah&st=0&or=10

1 AMD Athlon(tm)/Opteron(tm) CPU-model unknown 2 278,442.67 2,857.89 139,221.34 1,428.95
2 Intel(R) Xeon(R) CPU X5350 @ 2.66GHz 2 68,561.65 2,256.93 34,280.82 1,128.46
3 AMD Athlon(tm) 64 FX-70 Processor 2 39,195.18 1,984.78 19,597.59 992.39
4 Intel(R) Xeon(R) CPU X3220 @ 2.40GHz 4 689,678.66 3,450.13 172,419.67 862.53
5 Intel(R) Xeon(R) CPU E5335 @ 2.00GHz 5 106,276.05 4,133.46 21,255.21 826.69

[w0mbat]

Even if this K10 works @3GHz the results are amazing. But if he works @2,6GHz oder even @2,1GHz this results are like from a other world!

New Opteron @ ? (für die Rechnung werden übertriebene 3GHz angenommen)
1457 / 3000 = 0.4856
1457 / 2600 = 0,5603
1457 / 2100 = 0,6938

X6800 @ 3,784GHz
1503 / 3784 = 0.3971

A64 FX-60 @ 2,94GHz
755 / 2940 = 0.2568

[informal]

If it is K10(and the dual core variant) it is probably working at lower clock or the same as the release clocks will be(ie 2.1-2.3 GHz).
Having said that,the results are really amazing.
Imagine the Quad Core Barcelona and the damage it will produce

[trakslacker]

Is there really anything reliable to indicate that this is in fact Barcelona?

[biohead]

First solid Barcelona numbers tip up

Maybe it's K10 dual core variant,maybe it isn't.

Dude. :/

[w0mbat]

This cant be a X2 because his score is too high. Even a 3GHz+ X2 wont get a score like this. Show me a AMD CPU which is faster than a 3,78GHZ X6800

[Ubermann]

Is it possible to hack it so it shows "unknown" ?
And why is one of the es or whatever running seti ?

[LowRun]

Is there really anything reliable to indicate that this is in fact Barcelona?

Nope, nothing reliable there i'm afraid.

[w0mbat]

Of cause its possible to hack this - its always. And i dont know why some guy with a ES works on seti@home. Maybe because only this stopps him from benching

[Order]

Interesting and cool if true.

[Tenro]

Of cause its possible to hack this - its always. And i dont know why some guy with a ES works on seti@home. Maybe because only this stopps him from benching

Who knows?... only reason I can think of is that seti@home is an "ok" load tester, there are better options out there though. It's probably not real, but if it is... wow!!

[K404]

Ach..this could be wrong in a bazillion ways, but I still posted this info on EP thanks!

[SpikeP]

Well interesting news. This is the cpu i have been waiting for results from. i am still on a athlon x2 4400+ @ 2.7ghz and 2gb ddr ram. want to do a big cpu upgrade in the second half of this year, and held off the conroe. I hope this info is true.

Would like to see more though

- Paul

[xB0n3s]

I'm not so much into seti, but is it any faster than a highly overclocked or even stock Kentsfield Would be nice if there were any performance gain, but I guess we have to wait and see it for ourselfs.

[w0mbat]

But a 4GHz Brisbane wont be faster than a 3,78GHz X6800. And BOINC would show unknown AMD Athlon 64 X2 CPU.

[DTU_XaVier]

Just to clear up... Is the 4x4-platform recognized by SETI?? Otherwise it could make sense that SETI might only see two cores (one of the CPUs), while recieving work-output from four...

Best Regards

[[XC] leviathan18]

that is a k10 working right there

[w0mbat]

Its not SETI, its BOINC. And if u read the stats u will see some FX7x CPUs. So they can recognize QuadFX.

@brentpresley: I know u dont believe in my screen, but now u can see that im not the only one reporting at first from a dualcore k10.

[informal]

Just to clear up a couple of things:

1.BOINC DOES recognize QuadFX paltform as FX70,72 and 74.And it does show the number of cores correctly(you can see this as looking at the top number results table,FX70 is 3rd).So this is not 4x4 system with 2 cores not shown properly.

2.We do know this is NOT the Brisbane or any regular dual core K8(be it Athlon or Opteron) since it would need to work around 4.4 Ghz to outpace 3.78Ghz e6800 which is second place in top perf. table for dual cores.

3.We can speculate if this is true machine or not,but the chances are this IS the true AMD based machine with dual core chip inside(the number of cores would always be recognized correctly,no matter if the family name of the CPU is unknown)

4.Supercooling of AMD is not a fun sport since a cold bug,so i guess chances are very small this is some highly clocked and super-cooled cpu!Unless some one got a G2 SiGe Brisbane with no cold bug if such a cpu even exists(slim chances)

In the end a little joke .This could be a RHT K10 QuadCore in its hybrid dual core mode .After all,x-86 secret article did mention post K8 architecture for the supposed RHT and the K10 is post K8 arch

Cheers

[informal]

Oh no. PLEASE don't start the RHT rumor again. I thought we were past that.

Cmon you saw i said it was a joke .Little amusement won't hurt anyone
btw do you agree on my thoughts about the clock and possible Brisbane supercooled chip?

[Omastar]

Shouldn't there be a forum like 'Xtreme Speculation' for this type of thing?

[w0mbat]

This is not a single AMD Opteron, this is a dual socket plattform with 2 AMD Opterons. Like the 2nd place isnt a single Xeon X5350 - there are two Xeon X5350 on a dual socket plattform. So this can be a K10 quad or a K10 dual - we dont know - becaus u can only see the used CPUs, not the cores.

[informal]

Yes. Coldbug is a BEAR on the AMD chips.

I was noticing something on the BIONIC ranking though - the GHz numbers are what are pulled from the microcode table and don't take overclocking into account at all. So that makes this a bit more vague that I originally thought.

Are the top 2 chips OC vs. OC, stock vs. OC, stock vs. stock? On that we just don't know.

Also, does SETI use primarily Int, FPU, or SSE code? We have heard various rumors on how well K10 performs on SSE/FPU code but very little on Int.

Thoughts? Speculation (opening a CAN of worms there)?

Yeah,unfortunately a can of worms .But we can assume it uses FPU/SSE since it is the most upgraded part of the chip.
I think that the chips at first two places are probably not running stock.
Also note the weird number of CPUs at 7th place.Some faculty/institute is donating a lot of C2D love to BOINC .

[K404]

Whos DresdenBoy?

[w0mbat]

I think SETI@Home uses primary FPU. And u are right with the clocks. But i think the K10 isnt clocked faster than 2,66GHz cause now i thinks its a K10 quad.