First solid Barcelona numbers tip up

Quote:

---

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

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

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 :fact:

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

First solid Barcelona numbers tip up

Quote:

---

Originally Posted by informal

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

---

Dude. :/

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 :D

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

Quote:

---

Originally Posted by trakslacker

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

---

Nope, nothing reliable there i'm afraid.

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 :)

Interesting and cool if true.

Quote:

---

Originally Posted by 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 :)

---

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!!

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

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:D

- Paul

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

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

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... :confused: :fact:

Best Regards :toast:

that is a k10 working right there

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.

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 :D .After all,x-86 secret article did mention post K8 architecture for the supposed RHT and the K10 is post K8 arch :D

Cheers

Quote:

---

Originally Posted by brentpresley

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 :D
btw do you agree on my thoughts about the clock and possible Brisbane supercooled chip?

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

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.

Quote:

---

Originally Posted by brentpresley

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 :).

Whos DresdenBoy?

:)

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.

hopefulyl we will see a k10 and dual r600 setup from the AMD area of cebit. That would rock:slobber:

- paul

Quote:

---

Originally Posted by brentpresley

All reports from AMD are that they are going to release the QC parts LONG before the DC parts.

---

Which reports? Everything I've seen from AMD about dual cores indicates they expect to release them shortly after quads.

Quote:

---

Originally Posted by brentpresley

Bottom line - this is hopeful, but we need A LOT more info.

---

Definitely. Does anyone here run SETI@Home? How easy is it to hack?

Its all about BOINC and not SETI@Home. I use Boinc, but i dont know how to crack it. I think it checks the system data to show the CPU. Thats why u always read Intel(r) Core 2 Duo(r) CPU balblabal @ default clock.

Meaningless. You can run SETI@Home for 30 minutes on QX6700 per day. Or you can run SETI@Home for 24 hours per day on Pentium MMX. You will get more average credits for Pentium MMX.
BONIC Average credits is calculated by amount of the work was done per day.

But if u wanna be in the toplist u need to run BOINC 24/7. And the CPUs which we are comparing are in the top 10.

Quote:

---

Originally Posted by w0mbat

But if u wanna be in the toplist u need to run BOINC 24/7. And the CPUs which we are comparing are in the top 10.

---

Read this post:
http://www.aceshardware.com/forums/r...6607&forumid=1

Max credits per day:

AMD Athlon(tm)/Opteron(tm) CPU-model unknown 4,208
Intel(R) Core(tm)2 Quad CPU @ 2.66GHz 12,988

Quote:

---

Originally Posted by brentpresley

Goodness knows we've been waiting since last summer and the C2D release to see WHAT AMD's response would be.

---

Quote:

---

Originally Posted by brentpresley

3 months will seem like an eternity to the guys here wanting a cheap, high-powered DC chip when they are staring at great numbers from the guys with QC chips.

---

omg you're pathetic... you don't even know if that's a real K10, but since you can't bash it's performance you gotta bash something... If K10 quadcores come out at the end of Q2 and K10 dual cores at the beggining of Q3 it's that still 3 months appart? And how do you know the dual cores will be cheap? And how do you know that Quadcores won't be cheap? Cheap compared with what? High-powered DC?? Are you in panic-bash-mode? You got almost more posts in this thread then everybody else combine lol

i feel that same excitement growing in my stomach as when the C2D numbers were first trickling in....
everyone got their popcorn ready?:toast:

Quote:

---

Originally Posted by brentpresley

I guess it is a matter of what you consider a long timeframe.

http://www.hkepc.com/bbs/itnews.php?tid=679375

---

That chart says both quads and dual cores in Q3. There might be some quads in very late Q2, but I'd expect quads early-to-mid Q3 and dual cores mid-to-late Q3. I wouldn't be surprised if ES of both were out by now.

Quote:

---

Originally Posted by kl0012

Read this post:
http://www.aceshardware.com/forums/r...6607&forumid=1

Max credits per day:

AMD Athlon(tm)/Opteron(tm) CPU-model unknown 4,208
Intel(R) Core(tm)2 Quad CPU @ 2.66GHz 12,988

---

Thanks. Just wanted to post it here myself :)

Quote:

---

Originally Posted by kl0012

Meaningless. You can run SETI@Home for 30 minutes on QX6700 per day. Or you can run SETI@Home for 24 hours per day on Pentium MMX. You will get more average credits for Pentium MMX.
BONIC Average credits is calculated by amount of the work was done per day.

---

Exactly. Also I wonder how anyone concluded that the unkonw Athlon(tm)/Opteron(tm) is barcelona?
You can submit any CPU brand name/model string you want. For example: "IntelAMD", "AdvancedsMicrosDevices AMD", etc.
Just another BS from the_INQ. As always their goal is to make more traffic to their site.

Probably just a generic cheater.

Quote:

---

Originally Posted by brentpresley

uOpt, do you know if/how it is possible to submit fake scores to BIONIC?

---

Are you kidding? :)

I don't plan to give a list of options, but think over it. There are thousands of people on the net who's idea of a good time for their weekend is to write operating systems. How difficult will it be for them to encapsulate any process in any wrapper that changes "things"?

Isn't the Xeon X5350 Quad-core. That list shows it as 2 CPUs?

Quote:

---

Originally Posted by gOJDO

Exactly. Also I wonder how anyone concluded that the unkonw Athlon(tm)/Opteron(tm) is barcelona?
You can submit any CPU brand name/model string you want. For example: "IntelAMD", "AdvancedsMicrosDevices AMD", etc.
Just another BS from the_INQ. As always their goal is to make more traffic to their site.

---

What does this have to do with INQ ?

Quote:

---

Originally Posted by frankR

Isn't the Xeon X5350 Quad-core. That list shows it as 2 CPUs?

---

http://www.xtremesystems.org/forums/...5&postcount=27

Quote:

---

Originally Posted by Ubermann

What does this have to do with INQ ?

---

Nothing.

boinc is easily optimized. but i do hope this is a dual core chip and it is real because think of the quad core chips:slobber: and why is this guy running seti? he should join the xs wcg team:D :p: j/k

Quote:

---

Originally Posted by hipno650

boinc is easily optimized. but i do hope this is a dual core chip and it is real because think of the quad core chips:slobber: and why is this guy running seti? he should join the xs wcg team:D :p: j/k

---

Those are two independed systems with four cores each. See the post of Dresdenboy:
http://www.aceshardware.com/forums/r...6607&forumid=1
Actualy, nothing special.

Please be that real ;)
I'd like to change my sig :D

Quote:

---

Originally Posted by xoqolatl

Please be that real ;)
I'd like to change my sig :D

---

What's wrong with your sig :confused:

interesting ...

Quote:

---

Originally Posted by brentpresley

No, I'm not kidding. If you have to submit a completed workload along w/ an ID, then you would have to either harvest workloads or fake the ID of a VERY fast system (later seems a lot easier to me than the former).

---

I was running BOINC for over 2 years on my computers. Friend of mine found easy way of increasing scores for each computed packet. You started computing packet on FAST computer for which BOINC is evaluating cpu speed/memory/hdd/... and then computing it on SLOW machine (simple copy of folder). After few days when everything is 95% computed you're moving all data back on FAST computer and let him send everything back to BOINC servers.

Of course this gives huge gains only every couple of days and is very easy to trace on daily graph. I think this is what Intel guy from Dresdenboy link is doing to achieve 12000 points per day.

I guess I'm stupid or it's time to go to bed for, but what numers should I be comparing from this boinc chart??
I mean the numbers are allover he board, slower cpus with better scores, then the same cpu has bad scores, how can anybody take anything from this as a measure of anything??
________
ALASKA DISPENSARIES

Quote:

---

Originally Posted by VulgarHandle

i feel that same excitement growing in my stomach as when the C2D numbers were first trickling in....
everyone got their popcorn ready?:toast:

---

Yes, would you like some pudding?:D

Quote:

---

Originally Posted by Ubermann

What does this have to do with INQ ?

---

A definiton of pure the_INQ BS:

Quote:

---

First solid Barcelona numbers tip up

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.

---

Quote:

---

Originally Posted by Kingcarcas

Yes, would you like some pudding?:D

---

How can you have any pudding, if you don't eat your meat?

Quote:

---

Originally Posted by freeloader

How can you have any pudding, if you don't eat your meat?

---

We don't need no education! We need real results!

Now the killer numbers for AMD's Barcelona Core (albeit speculative and unreliable) is good news for server users, how does this apply to the Desktop market? Will this core be Socket 1207-only, or will there be an AM2/+/3 version? When will THAT come along? And are the AMD desktop processors coming out in Q3 based on that same K10 architecture?

EDIT: You! Yes, you behind the bike shed! Stand still, laddie!

there are dual-cores coming, with 2x512 L2 and 2mb L3 for AM2/+ as well as quad-core 4x512 L2

Quote:

---

By Mark Boslet
Of DOW JONES NEWSWIRES

SAN FRANCISCO (Dow Jones)--Advanced Micro Devices Inc. (AMD) cranked up the rhetoric Wednesday, claiming that its "quad-core" computer chip, code named Barcelona, outperforms a similar offering from rival Intel Corp. (INTC). (...)
At a briefing for analysts and the press, AMD of Sunnyvale, Calif., said Barcelona, which is expected in the third quarter, offers a 42% improvement over Intel's quad-core processor by one common measure of computing performance. (...)
At the briefing, AMD demonstrated a computer it claims is capable of "teraflop" performance. Once the realm of supercomputing, teraflop computers perform one trillion calculations in a second and are now being conceived as simpler, more modestly priced systems.
(...)
To achieve teraflop speeds, AMD's system uses two R600 graphics cards and two dual-core Opteron chips.
Nathan Brookwood, research fellow at Insight 64, a Saratoga, Calif., research firm, said the two teraflop computers were being designed for different purposes. Intel's effort is a research project while AMD's is an attempt at a commercial implementation.
(...)
-By Mark Boslet, Dow Jones Newswires; 690-496-1366; mark.boslet@dowjones.com
(END) Dow Jones Newswires

---

AMD will be string back more powerful than ever ;)

Perhaps the unknown AMD systems is also using a couple of GPU's for Seti/Bionics applications to get that high score. Who the heck knows ... :D

Quote:

---

Originally Posted by Flambo

Perhaps the unknown AMD systems is also using a couple of GPU's for Seti/Bionics applications to get that high score. Who the heck knows ... :D

---

As far as I know BOINC can't use GPGPU (yet :) ).

Quote:

---

Originally Posted by Lightman

As far as I know BOINC can't use GPGPU (yet :) ).

---

true, fortunately AMD is working with a couple dozen projects and getting them to support GPU coprocessors.