AMD Bulldozer Engineering Sample CPU Overclocked to 4.63GHz

[SonicIce]

1.29sec makes it very easy to conclude that its fake.

like others have said, the number is supposed to be between 10.029s and 19.929s. the spaces are whited out digits just like some of the other pieces of the image

 

[996GT2]

like others have said, the number is supposed to be between 10.029s and 19.929s. the spaces are whited out digits just like some of the other pieces of the image

Conclusion: Bulldozer's single-threaded performance is not that great

 

[psolord]

Conclusion: Bulldozer's single-threaded performance is not that great

Shouldn't we wait for more modern single cpu benchmarks, to make that conclusion?

 

[bridito]

My comment on:

http://seekingalpha.com/article/276697-amd-s-fortunes-rise-with-llano-but-risk-remains

has appeared after moderation but no reply yet. Should be interesting to see if any reply is made. It really is getting to the point where even the simplest statements made in the press can't be believed. We may be returning to the pre-media age where everything had to be touched, felt, or seen in order to be real!

 

[inf64]

Conclusion: Bulldozer's single-threaded performance is not that great

Wow you make that conclusion based on BD ES which actual performance numbers(that are questionable) are whited out while the benchmark itself(superPI) uses age old x87 math which in turn is severely depreciated since MS launched Windows X64 with AMD64 support.

 

[Ancalagon44]

So, its nearly the end of June, and we have no bulldozer, no bulldozer PR, and no bulldozer benchmarks. FAIL AMD!

 

[BonzaiDuck]

Let me posit this.

The current comparison of the Phenom X6 1055 and the i7-2600K is "more appropriate," even for comparing six apples to four oranges. But they're neck-and-neck in price and benchmark performance.

This Bulldozer 8-core processor -- waiting in an atrium for release to the congregation -- has 8 apples.

So I think it would be better yet to wait and see what happens with Socket-2011 and Ivy Bridge . . . .

 

[ed29a]

So, its nearly the end of June, and we have no bulldozer, no bulldozer PR, and no bulldozer benchmarks. FAIL AMD!

I am sure AMD execs are losing a lot of sleep over some people on some random forum complaining about the lack of dozer information. Afterall, we enthusiasts are their main source of income! Oh wait ...

 

[Ancalagon44]

I am sure AMD execs are losing a lot of sleep over some people on some random forum complaining about the lack of dozer information. Afterall, we enthusiasts are their main source of income! Oh wait ...

For a long time, we were pretty much their only source of income. Dell and HP wouldnt touch them due to Intel's rebates.

Even with things how they are, datacentre admins are going to make purchasing decisions based on benchmarks. Right now, were I a datacentre admin, I could see no good reason to hold off a server purchase for bulldozer.

But really, PR does also eventually reach the ears of the masses. Intel does a lot of PR, a lot of good PR, and AMD does none. Zip. Nada. Zilch. Who do you expect to sell more? We get one or two slides when Llano hit the stores, but thats it.

 

[Idontcare]

My comment on:

http://seekingalpha.com/article/276697-amd-s-fortunes-rise-with-llano-but-risk-remains

has appeared after moderation but no reply yet. Should be interesting to see if any reply is made. It really is getting to the point where even the simplest statements made in the press can't be believed. We may be returning to the pre-media age where everything had to be touched, felt, or seen in order to be real!

I think you need to develop a more sophisticated internal filtering and ranking system when it comes to sites you would classify as "the press" versus the sites you ought to be classifying as "some ignorant uneducated dude's personal opinion on topic xyz".

There is almost no distinction worth making nowadays between a forum post which generates ensuing debate within a thread, a rumor-mill speculation article which generates ensuing debate within the comment section of the article itself, and a blog which generates ensuing debate within the twitter space and so on.

Vet your sources and you'll develop a reliable network of contributors out there that you can rely on to paint a decent picture of reality for you. If you cast too wide a net and assume everyone knows everything they attempt to talk about then you are headed for a lot of disappointment.

Wow you make that conclusion based on BD ES which actual performance numbers(that are questionable) are whited out while the benchmark itself(superPI) uses age old x87 math which in turn is severely depreciated since MS launched Windows X64 with AMD64 support.

I keep reminding folks this same sentiment. Benchmarking beta-hardware, immature platforms, incomplete drivers, is not going to tell anyone anything of value.

Even sandy bridge was producing some rather mediocre results in those early leaks.

Even if the leaks are real, they are a static snapshot of a very dynamic and fast-changing storyline.

Look no further than GPU's and driver maturity, or the fact that we all can pretty much count on updating the BIOS on our mobo's at some point and you know that benches on pre-release hardware can possibly be expected to tell a reliable story about the performance of future retail products.

I think the case may be made that the leaks to speak to the minimum performance we can expect to see in the retail product (performance won't likely decline, unless TLB bug stuff happens), but I feel confident in saying the rule of thumb is that these leaks - if they are legit - merely paint a worst-case scenario of future performance.

I intentionally avoid looking at leaked benches on pre-release hardware, they can't possibly be expected to adequately characterize performance on release hardware and in the absence of pricing and power-consumption we can't possibly arrive at any sort of preliminary conclusion in terms of price/performance or performance/watt.

 

[Idontcare]

Even with things how they are, datacentre admins are going to make purchasing decisions based on benchmarks. Right now, were I a datacentre admin, I could see no good reason to hold off a server purchase for bulldozer.

Don't conflate the access to actionable information which professionals have versus the (lack of) access to actionable information which us lowly DIY'ers have.

Datacenter admins will have their account manager contacts in the respective companies and they will have access to information as needed and when needed to intersect their company's internal decision timelines.

The fact that AMD does not accomodate the DIY'ers in kind IS proof that we simply do not represent a materially significant portion of their sales and they do not see this changing in the near future (otherwise their actions would reflect such).

 

[bridito]

Let me posit this.

The current comparison of the Phenom X6 1055 and the i7-2600K is "more appropriate," even for comparing six apples to four oranges. But they're neck-and-neck in price and benchmark performance.

This Bulldozer 8-core processor -- waiting in an atrium for release to the congregation -- has 8 apples.

So I think it would be better yet to wait and see what happens with Socket-2011 and Ivy Bridge . . . .

Not really since the cores in BD are not really architecturally cores in the way they are in the Phenom X6 1055 and the i7 2600K. Therefore we're looking at apples vs. oranges vs. shoes.

Since we have nothing but blanked out and Photoshopped benchmarks for both SB-E and BD, there are very few statements that can be confidently made. One of them is that "according to information from Intel" SB-E quad core should exceed overall performance of the i7 2600K. Therefore we would have this scenario when the LGA2011 platforms finally debut in Q4.

X6 1055 < 2500K < 2600K < SB-E Quad < SB-E Hexa

I don't think anyone would argue that point (hey, but maybe they will.)

If (and it's a huge IF) this wild and crazy OBR benchmark assumption can be trusted, we may be looking at this scenario in Q4:

X6 1055 < BD < 2500K < 2600K < SB-E Quad < SB-E Hexa

I repeat that there is very little real information to base this or any other assumption on. However, if this were to be the case, IMHO IMHO IMHO it would translate to AMD effectively giving up any hope of regaining their long lost top performance crown. Even Trinity would not be able to re-establish their dominance as by then they would be coming up head to head with IB. That would be a horrible prospect for all involved as all computer enthusiasts have a vested interest in maintaining a vigorous and equitable competitive condition in CPU manufacturing.

 

[bridito]

I think you need to develop a more sophisticated internal filtering and ranking system when it comes to sites you would classify as "the press" versus the sites you ought to be classifying as "some ignorant uneducated dude's personal opinion on topic xyz".

There is almost no distinction worth making nowadays between a forum post which generates ensuing debate within a thread, a rumor-mill speculation article which generates ensuing debate within the comment section of the article itself, and a blog which generates ensuing debate within the twitter space and so on.

Vet your sources and you'll develop a reliable network of contributors out there that you can rely on to paint a decent picture of reality for you. If you cast too wide a net and assume everyone knows everything they attempt to talk about then you are headed for a lot of disappointment.

I agree completely that it is 100% wrong and outright foolish to assume that a Joe's CPU & Fishing Tackle Discussion Forum post by user AMDForeverTilIDie stating that BD at 1.8 GHz outperforms Haswell at 5.6 GHz by several orders of magnitude is gospel. I'm definitely not trying to support Seeking Alpha as I know little about them. However, given that they are supposed to be part of AOL and Huffpo and their About Us Page states that:

http://seekingalpha.com/page/about_us

"Seeking Alpha was named the Most Informative Website by Kiplinger's Magazine and has received Forbes' 'Best of the Web' Award."

And they also state that the site... "is primarily written by investors who describe their personal approach to stock picking and portfolio management, rather than by journalists."

I don't know if they qualify as "ignorant uneducated dudes"... but hey, maybe they do! I remember the good old days when you could read something in the NY Times or watch it on CBS News and although there could be some spin applied, at least you could trust the fundamental facts were accurate and checked by an editor prior to release. All of that does seem to have gone out the window, even with NYT, CBS, et al. Today's news items even from major media are riddled with errors from spelling and grammar to actual facts. I recently read a Wall St. Journal article which referred several times to Hugo Chavez' PROSTRATE cancer (where the cells metastisize on their knees, apparently). I'm starting to think like Rumsfeld: There are things we know that we don't know, there are things we...

 

[Idontcare]

I agree completely that it is 100&#37; wrong and outright foolish to assume that a Joe's CPU & Fishing Tackle Discussion Forum post by user AMDForeverTilIDie stating that BD at 1.8 GHz outperforms Haswell at 5.6 GHz by several orders of magnitude is gospel. I'm definitely not trying to support Seeking Alpha as I know little about them. However, given that they are supposed to be part of AOL and Huffpo and their About Us Page states that:

http://seekingalpha.com/page/about_us

"Seeking Alpha was named the Most Informative Website by Kiplinger's Magazine and has received Forbes' 'Best of the Web' Award."

And they also state that the site... "is primarily written by investors who describe their personal approach to stock picking and portfolio management, rather than by journalists."

I don't know if they qualify as "ignorant uneducated dudes"... but hey, maybe they do! I remember the good old days when you could read something in the NY Times or watch it on CBS News and although there could be some spin applied, at least you could trust the fundamental facts were accurate and checked by an editor prior to release. All of that does seem to have gone out the window, even with NYT, CBS, et al. Today's news items even from major media are riddled with errors from spelling and grammar to actual facts. I recently read a Wall St. Journal article which referred several times to Hugo Chavez' PROSTRATE cancer (where the cells metastisize on their knees, apparently). I'm starting to think like Rumsfeld: There are things we know that we don't know, there are things we...

My point was less about assessing the credibility of the establishment and more about establishing the credibility of the individual contributors.

Just as there is a range of credibility in forum posters within a given forum community, there will be a credibility range of journalists within any media organization.

Time and again the unwitting public has learned that internal scrutiny and editing by the editors is lackluster, your first line of defense is your own ability to be skeptical and non-gullible.

How many times has a journalist of brick-and-mortar news industry been caught (long after the publication date) of forging their "contacts" or photoshopping their images? Of claiming to be on the ground reporting in Beirut only to find out they are in Casablanca writing on their laptop in a 5-star hotel?

And its rarely the editor that catches the obvious, it is usually a member of the public, a reader, who takes stock of the article or its photo's and writes in to say "hey, wait a minute, did any of you actually read this before it was published?"...and then the newspaper or whatever institution issues a retraction and respond "oops, we should have picked up on this baloney long ago, our bad".

Quality can be found anywhere, there are quality bloggers, their are quality rumor-site authors, there are quality reporters in the media...but you proceed at your own peril if you substitute the credentials of the institution for those of the author of whatever you are reading IMO. That was all I was alluding to.

 

[Dribble]

It's wrong to say that sample is meaningless, or super pi is meaningless. Sure you can quantify it as not the finished product, but that's very different to dis-guarding all the results all together which tbh it sounds like people are doing mostly because they don't want them to be true.

Super pi is a perfectly acceptable test for single threaded (IPC) performance - there is a long history of super pi performance correlating to real usage performance.

The motherboards shouldn't be very early dev models. We already know BD was going to be released now and the MB makers designed boards for that aborted release. Hence we can assume that the boards being used now are probably release-ready.

BD itself is not being magically redesigned from that engineering chip used in the test to the release one. It's going through a respin, but that's it. Equally I highly doubt there are performance slowing locks or anything else in it - these chips exist partly for MB makers to use to make their boards, they wouldn't be happy being given intentionally crippled chips.

To back all this up month after month we've been hearing about IPC problems - sure every time we hear it that's just a rumour, but it's highly likely there isn't smoke without fire.

I agree that released BD will be different from that sample, but I suspect the main difference is it'll take less watts to run at that speed, not the actual peformance in super pi at that speed.

 

[bridito]

Time and again the unwitting public has learned that internal scrutiny and editing by the editors is lackluster, your first line of defense is your own ability to be skeptical and non-gullible.

Truer words were never spoken. I fully agree. Bravo! :thumbsup:

To back all this up month after month we've been hearing about IPC problems - sure every time we hear it that's just a rumour, but it's highly likely there isn't smoke without fire.

I agree that released BD will be different from that sample, but I suspect the main difference is it'll take less watts to run at that speed, not the actual peformance in super pi at that speed.

Therefore are you saying that:

X6 1055 < BD < 2500K < 2600K < SB-E Quad < SB-E Hexa

could actually be a reality?

 

[996GT2]

Therefore are you saying that:

X6 1055 < BD < 2500K < 2600K < SB-E Quad < SB-E Hexa

could actually be a reality?

Come on now, would you really be surprised that Bulldozer is not faster than Sandy Bridge? That's simply what I've been expecting all along, which is why I didn't even bother waiting for BD and just upgraded straight to Sandy Bridge for my build earlier this year.

 

[guskline]

Come on now, would you really be surprised that Bulldozer is not faster than Sandy Bridge? That's simply what I've been expecting all along, which is why I didn't even bother waiting for BD and just upgraded straight to Sandy Bridge for my build earlier this year.

Did the same thing.

I'm only one person, but my $$ went to Intel, not AMD, for the first time since the early 90's.

 

[sm625]

There is probably not a single BD out there in the wild that does not have its L2 cache disabled. Yet we have all these turkeys here on these forums that "just know" a 2500k is faster. That is a fine display of ignorance.

 

[OCGuy]

There is probably not a single BD out there in the wild that does not have its L2 cache disabled. Yet we have all these turkeys here on these forums that "just know" a 2500k is faster. That is a fine display of ignorance.

Pretending to know anything firm about BD performance is ignorant, yes. But so is believing that the BD delay is due to AMD needing to use the fabs to fill Llano orders. That seems to be a commonly parroted idea that originated from a AMD marketing press release.

We are all sitting around discussing rumours and fake slides, as well as possibly real leaked ES results...but no way of knowing one from the other.

 

[996GT2]

There is probably not a single BD out there in the wild that does not have its L2 cache disabled. Yet we have all these turkeys here on these forums that "just know" a 2500k is faster. That is a fine display of ignorance.

Betcha $5 that BD does not beat Sandy Bridge whenever it comes out

 

[bridito]

Come on now, would you really be surprised that Bulldozer is not faster than Sandy Bridge? That's simply what I've been expecting all along, which is why I didn't even bother waiting for BD and just upgraded straight to Sandy Bridge for my build earlier this year.

By some of the comments I've been reading from AMD enthusiasts it seemed to them that it was going to be:

X6 1055 < 2500K < 2600K < SB-E Quad < SB-E Hexa < Cray < Skynet < BD < God

 

[Phynaz]

There is probably not a single BD out there in the wild that does not have its L2 cache disabled.

First I've heard on this. Why is AMD disabling L2? I assume this would also disable L3.

 

[996GT2]

By some of the comments I've been reading from AMD enthusiasts it seemed to them that it was going to be:

X6 1055 < 2500K < 2600K < SB-E Quad < SB-E Hexa < Cray < Skynet < BD < God

Yeah they said the same thing right before the delayed release of Phenom

 

[Idontcare]

Come on now, would you really be surprised that Bulldozer is not faster than Sandy Bridge? That's simply what I've been expecting all along, which is why I didn't even bother waiting for BD and just upgraded straight to Sandy Bridge for my build earlier this year.

Considering the resources Intel has plowed into developing not only Sandy Bridge's architecture but also the 32nm process tech that enables it's clockspeeds and power-consumption...the only thing that we should expect to be able to beat Sandy Bridge is Ivy Bridge.

Miracles can happen, but Sandy Bridge is no Pentium 4 Netburst step-back situation. Miracle denied IMO.