Bulldozer ES benchmark is out!

[Black96ws6]

This is a pretty good re-cap of where we are at.
AMD Delays Bulldozer Citing Performance Issues

That article does not paint a good picture of Bulldozer performance AT ALL

 

[Nemesis 1]

http://www.youtube.com/watch?v=mr7kr4kimeM&feature=related

 

[Vesku]

I think it would take some very reliable evidence for me to start thinking AMD is going to release a bulldozer that performs worse than their K10+ Llano chips. Now if you were to say the IPC was actually going to be 10-15% lower than Nehalem rather than close to parity, that I could get pessimistic about. Would add to the narrative of AMD trying to get better clocks out of their 32nm, they may need that 4.2GHz turbo single threaded just to match 3.6GHz Sandybridge single threaded.

 

[Black96ws6]

I think it would take some very reliable evidence for me to start thinking AMD is going to release a bulldozer that performs worse than their K10+ Llano chips. Now if you were to say the IPC was actually going to be 10-15% lower than Nehalem rather than close to parity, that I could get pessimistic about. Would add to the narrative of AMD trying to get better clocks out of their 32nm, they may need that 4.2GHz turbo single threaded just to match 3.6GHz Sandybridge single threaded.

Well if the above article is correct, it sounds like that was the case for the B0 stepping, and they're trying to fix it as well as get the clock speeds up...hence the delay from June to September (3 months)...because it sounds like they did want to release in June but just couldn't because they weren't competitive enough at that point...

 

[sm625]

You can easily see that superPi scores improve with newer, better cpus. It just looks like a really straightforward single threaded benchmark. I dont get why so many are trashing it. To change benchmarks now we lose our frame of reference. There is no compelling reason to do so at this time.

As for cinebench, what that video shows us is that he is indeed running an 8 core processor. And it is not a bad score at all considering the cache is disabled. I would expect much worse. I really dont get how it can perform as well as it does with a broken L2&L3.

 

[Drakula]

Reading the review or leaked benchmark number, whatever you call it, from the link that OP posted, and was wondering why is Cinebench showing that GFX Board is a Radeon 2400 Pro?

 

[mrcmtl]

Why is everyone so sure that those ES chips are actually running at the frequencies showed? If so, then hell, a 5.4Ghz llano chips sounds promising.

 

[Topweasel]

You can easily see that superPi scores improve with newer, better cpus. It just looks like a really straightforward single threaded benchmark. I dont get why so many are trashing it. To change benchmarks now we lose our frame of reference. There is no compelling reason to do so at this time.

Easy if a benchmark is not an example of some kind of real life work load, then it can no longer be used as frame of reference.

I wouldn't suggest we base our GPU reviews on how well GPU runs Quake 1. For multiple reason, sure its a graphics based application, and sure all the other hardware is the same. But a 5 Trillion FPS result and 3 Trillion FPS result is meaningless test when the real question is. Can it run Crysis? Guess what in those tests and in todays benchmarks, lots of times its the card that loses the unrealistic and 0 impact, low res, old engine test, that wins the normal resolution, high IQ, tests.

Basing any performance information on a test that used to take several minutes and is now down to a handful of seconds, pretending that an extra handful of seconds means half the performance and slow CPU, is pretty useless.

A long time ago there was a steady move away from synthetic tests and fake workloads on video cards, because it just mean that driver developers were spending more time trying to cheat then improve stability and quality. I would like to see that happen on CPU's. lets stop trying to extrapolate real life performance characteristics out of a outdated unrealistic test that's become so small and quick even the smallest of I/O hiccup is going to double the score.

 

[Makaveli]

Sure, but it has 8 cores...you'd think that'd make up for the lower IPC...

lol what??

Why would that make a difference when most apps can't even use 4 cores?

 

[3DVagabond]

lol what??

Why would that make a difference when most apps can't even use 4 cores?

If you don't have apps that can use more than 4 cores, what's the point of considering an 8C processor?

I've said this before, but some people do more than play games on their computers. If you don't, then it's very likely that BD is not the right processor for you. The 2500K is probably the optimum processor to spend your money on. Spend the money saved on a better GC or Monitor.

 

[radaja]

anyone see this yet?maybe a B1 BD?

http://obrovsky.blogspot.com/

 

[dx11101]

If this chip was an intel killer they would be screaming it from the tops of the mountains by now. This is reminding me a lot of the phenom launch where benchmarks were top secret until the last second. All the rumores then turned out to be a bunch of BS and then everyone was just disappointed in the end. Especially because of the whole errata in the l3 cache for the 9500 and 9600 phenoms. Plus AMD launched it at clocks way lower then they claimed they would.

 

[RoyG]

If this chip was an intel killer they would be screaming it from the tops of the mountains by now. This is reminding me a lot of the phenom launch where benchmarks were top secret until the last second. All the rumores then turned out to be a bunch of BS and then everyone was just disappointed in the end. Especially because of the whole errata in the l3 cache for the 9500 and 9600 phenoms. Plus AMD launched it at clocks way lower then they claimed they would.

+1

 

[Vesku]

That's the main reason for the delay I believe, they want to squeeze some more out of GloFo 32nm.

 

[StinkyPinky]

If this chip was an intel killer they would be screaming it from the tops of the mountains by now. This is reminding me a lot of the phenom launch where benchmarks were top secret until the last second. All the rumores then turned out to be a bunch of BS and then everyone was just disappointed in the end. Especially because of the whole errata in the l3 cache for the 9500 and 9600 phenoms. Plus AMD launched it at clocks way lower then they claimed they would.

Exactly. They should rename Bulldozer to Fido because I think it's going to be a dog.

Let's be honest. It's been delayed because of sub-par performance. If they were happy with it they'd be leaking benchmarks by now to generate hype.

 

[piesquared]

Generate hype. Can you think of a product that has more hype than this one? What's the point in showing their hand? If the products are good, they'll sell just as many in 2 months, as they would if they let the cat out of the bag now. For those that say intel already knows how it performs, didn't they recently release a leak to the press stating that depending on how Bulldozer performs, they could adjust price. That seems to indicate they don't have concrete numbers either. So again, what's the point? Once all the individuals saying they've had enough and are just going to run right out right now and buy an intel product because they aren't going to wait any longer, get weeded out from the individuals that are actually going to run right out and buy an intel product because they aren't going to wait any longer, I doubt they'll lose a lot of sales.

 

[996GT2]

Generate hype. Can you think of a product that has more hype than this one? What's the point in showing their hand? If the products are good, they'll sell just as many in 2 months, as they would if they let the cat out of the bag now. For those that say intel already knows how it performs, didn't they recently release a leak to the press stating that depending on how Bulldozer performs, they could adjust price. That seems to indicate they don't have concrete numbers either. So again, what's the point? Once all the individuals saying they've had enough and are just going to run right out right now and buy an intel product because they aren't going to wait any longer, get weeded out from the individuals that are actually going to run right out and buy an intel product because they aren't going to wait any longer, I doubt they'll lose a lot of sales.

It is what it is. When Bulldozer comes out a full 9 months after Sandy Bridge, I have little doubt that it will still be slower on a core by core basis. Maybe it will have a bit of an advantage in very heavily multithreaded apps, but for most games and apps Sandy Bridge will still have the preformance lead.

Not much of a reward for loyal AMD fans waiting 9 months after SB, eh?

 

[3DVagabond]

It is what it is. When Bulldozer comes out a full 9 months after Sandy Bridge, I have little doubt that it will still be slower on a core by core basis. Maybe it will have a bit of an advantage in very heavily multithreaded apps, but for most games and apps Sandy Bridge will still have the preformance lead.

Not much of a reward for loyal AMD fans waiting 9 months after SB, eh?

Unless you can actually use 8 cores. Then it just sucks to have to wait.

 

[Arg Clin]

If this chip was an intel killer they would be screaming it from the tops of the mountains by now. This is reminding me a lot of the phenom launch where benchmarks were top secret until the last second. All the rumores then turned out to be a bunch of BS and then everyone was just disappointed in the end. Especially because of the whole errata in the l3 cache for the 9500 and 9600 phenoms. Plus AMD launched it at clocks way lower then they claimed they would.

I beg to differ. They chip could be preforming really well already, but unless AMD is also ready to deliver on volume production now it would be downright foolish to get everyone excited about a vaporware chip and give Intel the time to put the afterburner on IB.

AMD succeded with the K7 because they took Intel completely by surprise.

Given the lack of balance in resources between Intel and AMD, this is really the only AMD strategy that would make sense.

If BD was not competitive with intels offerings, at least in the mid end, AMD would have abandoned it already. It would be throwing good money after bad, to keep developing on an inferior product. We all know there are no money to be made in the low end segment.

 

[Idontcare]

If BD is as bad as some are thinking I would have to believe that AMD would have redesigned, or scrapped it, long ago.

Is that what Intel did with Prescott? Or did they bring it to market despite all its lackings?

 

[3DVagabond]

Is that what Intel did with Prescott? Or did they bring it to market despite all its lackings?

Was Prescott slower than what they already had?

 

[Idontcare]

Was Prescott slower than what they already had?

I can't tell if you are being facetious or if you genuinely are unawares of the history of Prescott...

http://www.anandtech.com/show/1230

90nm Prescott was slower (less IPC and did not clock higher), hotter, and higher power-consuming than its older 130nm Northwood sibling.

And yet Intel brought it to market anyways, knowing all this to be true well in advance of launch day and they continued to sell them as their leading product until the launch of Core 2 Duo >2yrs later.

 

[3DVagabond]

I can't tell if you are being facetious or if you genuinely are unawares of the history of Prescott...

http://www.anandtech.com/show/1230

90nm Prescott was slower (less IPC and did not clock higher), hotter, and higher power-consuming than its older 130nm Northwood sibling.

And yet Intel brought it to market anyways, knowing all this to be true well in advance of launch day and they continued to sell them as their leading product until the launch of Core 2 Duo >2yrs later.

Well, then I guess there's precedent. I'll be surprised if history repeats itself. We'll see soon enough, I suppose.

(Honestly had no idea. I haven't been into hardware for that long.)

 

[classy]

With AMD its the same story. Actually there is nothing wrong with Bulldozer except one thing, its late. We are looking at a processor that is about 2 years late to the party, so its performance would have probably been industry leading. But because of poor execution bringing it to the market, now it has to face updated cpus from Intel. AMD's ability to design is as good if not better than Intel in some regards, but they are very poor at bringing a product to the market. Its a shame really, because if they would have worked on improving their product execution with the dominate success they had with K7, they would be challenging to be a market leader. They have very poor management at that company.

 

[Idontcare]

We are looking at a processor that is about 2 years late to the party, so its performance would have probably been industry leading.

If this came out two years ago then it would have been on 45nm.

I don't see how making it hotter and slower would have made it any more competitive "in its day".

How did that work out for Phenom? What Phenom needed, and eventually got, was a shrink from 65nm to 45nm. It did not need to be released 2yrs earlier on 90nm.

Now if what you meant to say is that the BD we are going to see released this summer (i.e. the 32nm one) should have come out two years ago then yes, I totally agree. A 32nm-based BD in 2009 would have rocked.