Here's a little teaser....

[s7e9h3n]

can the Tyans oc at all ?

Not with the AMD-modified bios that it was running.

[Antinomy]

Step by step.
doompc, my nick is a little different from what you have written
There aren't English versions of those articles, only of the mentioned one. Yeah, the timings is one of the most popular of the articles of mine. They are on my home page on Overclockers.ru. But the second one is an article of Laboratory of the site, so I better get a permission on this stuff. If you're interested in this, better discuss it here.

tictac The HT link and the L3 performance aren't connected so you won't speed up the L3 by changing the HT multiplier. But we can speed it up by o\cing the CPU and\or Northbridge. I'm not sure that the BIOS in the near future will allow to overclock North bridge separately, but I'll look for this in the BIOS developers guide.

dave_graham - I don't wanna make you upset, but there's no way of BSEL mode on this CPU. The reason is simple - the LDT frequency is 200MHz. And its final frequency depends only from the HT multiplier. We can change it with the BIOS mod, I suppose, but FWIW? It won't give even a couple of percents, cause the speed isn't limited by the HT performance.

[grunge100]

Will we see a little more performance with the K10 cpu's with a later revision? Scientia over at amdzone thinks the SSE performance is below average considering it has been widened to 128 bit.

[tsuehpsyde]

Supermicro claims the problems we're having were fixed, so we're in contact with them now trying to find the problems and iron out the bugs. I'll keep everyone updated as I know more.

[KTE]

Cheers tsuehpsyde

I want to show you a little political babble - how all these media scores are just supporting of "who can deceive the most".

Taking 1, 2 and 3, 4 of the last submitted best scores for a K8 3GHz Opteron and a C2 2.93GHz X6800, averaging them both, and you find in SPECint_rate 2006 using different compilers, clock per clock basis, the C2D was 15.5% ahead.

Now there's more to add there: X6800 had a 8% lead in December 06 results, but due to their compiler and flags used, August 07 results are double % higher, with the exact same hardware. Opteron didn't.

SPECfp_rate 2006 was 16.4% lead by X6800 with the latest August scores and 6.6% with the December 06 ones, clock per clock basis.

Wotdodatshow?

Well, that you just cannot compare server application performance figures to desktop, for one, and secondly, in the server market K8 per clock basis wasn't at all far behind the Core 2.

If K10 Opty is topping at 2.5GHz, FX should be around 2.6GHz at least, X4 at 2.4GHz and X2 at 2.8GHz. That's my minimum expectation for top December releases.. if that's when they do come.

[akaBruno]

Hey, good stuff. The intel smell permeates it once again.

Since the 6400+ is overclocking to 3.6, I see no reason that the new cores will do at least that. And that's just a start.

I know it's all guess work at this point. And it may be nothing more than wishful thinking but... It won't be as bad as what the intel trolls are preaching.

[funnyperson1]

Hey, good stuff. The intel smell permeates it once again.

Since the 6400+ is overclocking to 3.6, I see no reason that the new cores will do at least that. And that's just a start.

I know it's all guess work at this point. And it may be nothing more than wishful thinking but... It won't be as bad as what the intel trolls are preaching.

Dude, AMD wouldn't be releasing 2.0GHz CPUs if they could hit 3.6 on air without heavily cherrypicking.

[Sup3rman]

Hey, good stuff. The intel smell permeates it once again.

Since the 6400+ is overclocking to 3.6, I see no reason that the new cores will do at least that. And that's just a start.

I know it's all guess work at this point. And it may be nothing more than wishful thinking but... It won't be as bad as what the intel trolls are preaching.

6400+ = 90nm?

[Sparky]

6400+ = 90nm?

Yes it is, and being more mature it is reaching higher speeds currently than 65nm. But I've been hearing of brisbanes hitting 3GHz + so I wouldn't be surprised to see stock ~3GHz 65nm K10's in the semi-near future (like 3-6 months or so) for their dual cores at least, quads probably won't hit it as fast.

Just speculating here, but I think we can all agree that AMD's 65nm just needs to mature more before we start seeing higher stock speeds. It's getting there! Just not fast enough it seems

[stealth17]

6400+ = 90nm?

so what? It's a whole different architect and considering everything it has on it I could see it not clocking as high. Take the L3 for example, it's just another thing to be a weak link in the chain while overclocking.

[PhilDoc]

so what? It's a whole different architect and considering everything it has on it I could see it not clocking as high. Take the L3 for example, it's just another thing to be a weak link in the chain while overclocking.

I think that's what he was suggeting. That it would get that, well at least not yet.

[KTE]

Can someone point me to an adequate link or source explaining why these are fake?

http://img296.imageshack.us/my.php?i...celona1ph6.jpg
http://img235.imageshack.us/my.php?i...celona2rs4.jpg
http://img513.imageshack.us/my.php?i...celona3ua6.jpg

That checksum is valid BTW.

1x Opteron 2210 1.8GHz x64
Cinebench10 - 1710/6219: http://img222.imageshack.us/img222/3...217e16dzg4.jpg
Fritz chess, CPUMark, SPI 1M, Geekbench, wPrime: http://img300.imageshack.us/img300/4...21a0b59iw9.jpg

[Bail_w]

Core = 1 ? thread = 1 ?

[justapost]

1x Opteron 2210 1.8GHz x64[/U][/B]
Cinebench10 - 1710/6219:
Fritz chess, CPUMark, SPI 1M, Geekbench, wPrime:

These benchmarks where run on 2 cores.

[KTE]

Yeah I posted your benches once before and mentioned that - I know they're real

But what about the K10 ones above them? If the pics are all valid, so is the checksum, and we know CPUZ erred before 8/10 with Agena details at least when they were posted, how are those any more fake than anything else anyone posts? If not, why are they that much different and extremely fast seeming compared to rest of guys here?

Comeon guys I'm trying to get to the boot of the truth here

[justapost]

Yeah I posted your benches once before and mentioned that - I know they're real

You wrote 1x2110 but it where 2x2210. That's why i noted.

But what about the K10 ones above them? If the pics are all valid, so is the checksum, and we know CPUZ erred before 8/10 with Agena details at least when they were posted, how are those any more fake than anything else anyone posts? If not, why are they that much different and extremely fast seeming compared to rest of guys here?

Comeon guys I'm trying to get to the boot of the truth here

You can use a valid cpu-z screenshot in your favourit image manipulation software.
I'm abit curious about gary's 26s-superpi comments on anandtech. I expected superpi around 20s-26s.

[KTE]

OK. Yeah (Gary comments), thats why I posted that as well... as it comes closer than what we've seen here or in japan yet. Have you seen the angles of the shots (13sec)? That's not image manipulation, those are camera pics very clearly.

The only doubtfulness I have of them is, whether the CPU-Z shot was the image manipulated and inserted onto the screen before taking a camera pic. I can't see a valid reason otherwise apart from having to apply the same logic to everyone else whose posted SPI times as well - and saying they're all fake unless they make me a video.

This is what Gary said that you're referencing:

If you compare a B00 chip from May to a B02 today, there is a significant difference in performance in all areas (26 seconds in SuperPI 1m for one) and my comments represent observations of what has occurred over this time period. We have final silicon now and results will be posted in the near future. My observations today are different than they were two weeks ago and as the platform matures they will change again.

Once we see the HT 3.0 capable chipsets and Phenom cores mature then we will have an even better indication of the performance of this core design in the consumer market but for now the initial release is Barcelona in the enterprise market.

Doesn't it seem to you that he's saying 26sec is the "difference" between the B0 and B2 chips rather than the "actual time" of the run?

Depending upon the application, Barcelona clock for clock is equal to Clovertown or just a tad faster in some areas and in others it is 20%+ (heavy emphasis on "plus" until Monday, ) faster. I have to say that final silicon is looking really nice at this point, especially in memory sensitive applications where this processor shines, but core speeds need to come up in a hurry to compete with Tigerton in general server applications.

The memory bandwidth numbers we noticed are a little surprising considering the results in certain memory intensive applications. The latencies are much improved over the last core stepping we looked at as is pure throughput which makes you think twice about the relationship between bandwidth and performance in general from this CPU.

And there's more that he said back then.

[doompc]

KTE, those are fake, SP33DFR34K did them:
http://www.xtremesystems.org/forums/...157136&page=21

edit:

Step by step.
doompc, my nick is a little different from what you have written

Sorry about that, Antinomy.

[justapost]

Doesn't it seem to you that he's saying 26sec is the "difference" between the B0 and B2 chips rather than the "actual time" of the run?
And there's more that he said back then.

You are right he's talking about the difference between B0's and B2's.
But we only have seen benchmarks from B1's or BA'2, so we don't know how bad B0 really was (say 60s super-pi).

[hipro5]

I haven't writen anything in this thead (I think so ).......So.....We are ALL just .....talking and talking and talking till now?.....Any real GOOD benchies out there yet or should we wait till X-Mas of.....what year we say?....

[Xtasy]

can somebody run the barcelona with 3dmark06 ?

[Motiv]

I haven't writen anything in this thead (I think so ).......So.....We are ALL just .....talking and talking and talking till now?.....Any real GOOD benchies out there yet or should we wait till X-Mas of.....what year we say?....

all we have are these. Not exactly stirling stuff but better than nothing

http://www.xtremesystems.org/forums/...&postcount=516

http://www.xtremesystems.org/forums/...&postcount=521

http://www.xtremesystems.org/forums/...&postcount=527

[dinos22]

I haven't writen anything in this thead (I think so ).......So.....We are ALL just .....talking and talking and talking till now?.....Any real GOOD benchies out there yet or should we wait till X-Mas of.....what year we say?....

that's just mean

they can't even run 3d at stock settings let alone post *GOOD* benches heh

[hipro5]

that's just mean

they can't even run 3d at stock settings let alone post *GOOD* benches heh

You're mean....

[dave_graham]

good god guys....

a.) currently, the only boards that have shipped have included x8 PCIe slots with NO support for PCIe cards. in the case of the Asus DFNS-DRE board, that hardly qualifies as much of anything since it's definitely geared towards I/O cards, not video.
b.) Tyan has confirmed that early B2 stepping parts are working with bog-standard S2915s (of which i have 4) so, I should be able to test and see here within the next few weeks. (on BAs, not B2s)
c.) AMD has already stated that B2 processors will supercede the BA stepping upon launch in November. Channel Availability, if given the typical TTM of 6 weeks, will be the end of December/early January then.

we're doing this gratis while trying to balance out the other needs that customers who are actually BUYING systems need to have done. So, just give us room to operate.

the fact that AMD botched the intro on these processors is already established and repeating it (or inferring it) ad naseum in this thread will get your posts removed.

done.

dave