AMD cuts to the core with 'Bulldozer' Opterons

[Manicdan]

i predict that games in 2014 will still not be cpu limited, cause as things that use the cpu get more intense, they try to move it to the gpu, like physics did

[informal]

Stop this nonsense. You claim total performance doesn't matter in real world, with this as proof?!
But after that you claim that L3 has an invisible impact, and core count?
Let me get this straight. If I have my Phenom II running a bunch of games today clocked at 2GHz and 4.8GHz in two runs. Limited by my GPU I get the exact same FPS in both cases. Will both perform equally in the games realesed in 2014? Same core count, same cache, same everything except frequency.

What kind of a game would run the same @ 2Ghz and 4.8Ghz Deneb chip?? There is no such a game and if there was it would be extremely GPU bound(I can't emphasize the word extremely).Both cache and core count are important(again look at Agena and Deneb and look at C2D Vs C2Q in modern games). There is a point where games stop scaling with CPU clocks since the GPU (yes even 5970) starts to bottleneck and can't process enough data.This happens with both Deneb and Nehalem.

Also first you need to find and show me "a bunch of games today" that can run the same @ 2Ghz and 4.8Ghz .There are no such games as I mentioned previously above.Second it wouldn't mean the games of 2014 would run the same on those 2 different CPU clocks since the games would a) be hardly playable with your current GPU if you would stick with it b) scaling would stop somewhere in between those two frequencies if you choose to buy a new GPU.The only way a game from 2014 could scale perfectly with clock speeds from 2 to 4.8Ghz is if it was coded with awesome multi core support and it uses all available CPU resources to the maximum(highly unlikely).

I'm not to sure, we don't know how fast the L3 is. It could be enhanced much. I often see i7 with 3 times the bandwidth. Faster L3 could improve scaling.

L3 sharing policies are very different in Deneb/Thuban and Nehalem so you can't just compare the bandwidth like that.AMD uses victim(spill over cache) while intel uses inclusive.

[-Boris-]

What kind of a game would run the same @ 2Ghz and 4.8Ghz Deneb chip?? There is no such a game and if there was it would be extremely GPU bound(I can't emphasize the word extremely).Both cache and core count are important(again look at Agena and Deneb and look at C2D Vs C2Q in modern games). There is a point where games stop scaling with CPU clocks since the GPU (yes even 5970) starts to bottleneck and can't process enough data.This happens with both Deneb and Nehalem.

Also first you need to find and show me "a bunch of games today" that can run the same @ 2Ghz and 4.8Ghz .There are no such games as I mentioned previously above.Second it wouldn't mean the games of 2014 would run the same on those 2 different CPU clocks since the games would a) be hardly playable with your current GPU if you would stick with it b) scaling would stop somewhere in between those two frequencies if you choose to buy a new GPU.The only way a game from 2014 could scale perfectly with clock speeds from 2 to 4.8Ghz is if it was coded with awesome multi core support and it uses all available CPU resources to the maximum(highly unlikely).


L3 sharing policies are very different in Deneb/Thuban and Nehalem so you can't just compare the bandwidth like that.AMD uses victim(spill over cache) while intel uses inclusive.

I admit, I exaggerated a bit. Say a 2.8GHz Phenom II and a 4.8GHz one. Would they handle the games released in 2014 equally?


And I know there is differences between i7 and Phenom II caches. My point is we don't know if DB has 48-way, 64-way or 96-way caches. We don't know if they operate around 2GHz or if the are at core speed. We know nothing about the speed of communication between modules.

[informal]

I admit, I exaggerated a bit. Say a 2.8GHz Phenom II and a 4.8GHz one. Would they handle the games released in 2014 equally?


And I know there is differences between i7 and Phenom II caches. My point is we don't know if DB has 48-way, 64-way or 96-way caches. We don't know if they operate around 2GHz or if the are at core speed. We know nothing about the speed of communication between modules.

2.8Ghz to 4.8Ghz is 71% improvement and this brings very minor improvement in today's games.In future games I think the GPU based PhysX (not necessarily meaning the NV's approach,using just the term) will play much bigger role and then ,again,CPU will become even less of a factor than it is today.What will mater is a number of relatively fast cores(IMO 4) and cache(by this I mean,at least Penryn and Deneb class @ 2.8-3Ghz).With offloaded PhysX to GPU and massively shader heavy game engines,CPU will use additional cores to offload AI for example,among other things.Also we have to keep in mind that maybe games will use new AVX instruction set and this may play a role in how certain chip performs.

[tifosi]

Server Bulldozer is to desktop bulldozer what server MC is to Phenom II.

Personally, i believe Bulldozer will be much more capable than these numbers suggest. 4 ALUs, loads/stores per clock, shared FPU, 32nm, and totally new core probably capable of much higher performance/watt suggest that bulldozer will be very very good.


So I'm hoping I'm wrong here. Since I am already planning to throw my AM2+ system away and buy an AM3+ system next year.

Shanghai's desktop equivalent was Phenom II (Agena). Now Istanbul was the 6-core Server processor that AMD came out with later on and its desktop equivalent(re-engineered) is Thuban. Magny Cours is a different core and was not worked into desktop platforms.

So... what it means is, improvements from the core modifications/ improvements in MC have not really been witnessed in desktop arena, as the desktop chippery is still based on older Shanghai and Istanbul equivalents. From that you could deduce that desktop chips should ideally see a really nice boost. Personally, i'd not be expecting anything less than 20-25 odd percent in single threaded apps, given increased cache and new instruction sets and well a hell of a lot more in multi-threaded scenario.

[Movieman]

Slightly OT but from what I see AMD with the MC's has got a excellent product, just need to get the clocks up.
A 12 core MC at 3000mhz would be a force to reckon with.

[-Boris-]

2.8Ghz to 4.8Ghz is 71% improvement and this brings very minor improvement in today's games.In future games I think the GPU based PhysX (not necessarily meaning the NV's approach,using just the term) will play much bigger role and then ,again,CPU will become even less of a factor than it is today.What will mater is a number of relatively fast cores(IMO 4) and cache(by this I mean,at least Penryn and Deneb class @ 2.8-3Ghz).With offloaded PhysX to GPU and massively shader heavy game engines,CPU will use additional cores to offload AI for example,among other things.Also we have to keep in mind that maybe games will use new AVX instruction set and this may play a role in how certain chip performs.

So you believe that there will be no practical difference between the two? You are very vague.

Shanghai's desktop equivalent was Phenom II (Agena). Now Istanbul was the 6-core Server processor that AMD came out with later on and its desktop equivalent(re-engineered) is Thuban. Magny Cours is a different core and was not worked into desktop platforms.

So... what it means is, improvements from the core modifications/ improvements in MC have not really been witnessed in desktop arena, as the desktop chippery is still based on older Shanghai and Istanbul equivalents. From that you could deduce that desktop chips should ideally see a really nice boost. Personally, i'd not be expecting anything less than 20-25 odd percent in single threaded apps, given increased cache and new instruction sets and well a hell of a lot more in multi-threaded scenario.

Thuban still is to MC what Zambezi is to Interlagos. They have the same relationships. Thuban is AM3 desktop, MC is two such dies for G34. Zambezi is AM3 desktop, Interlagos is two such dies for G34.

A 12 core MC at 3000mhz would be a force to reckon with.

Yeah, for heatsink manufacturers.

Honestly, I'm impressed with AMDs achievements with MCs power usage. But I think that 3GHz would be impossible on 45nm. It have to consume a third of what Phenom II 940 does per core. I think 12 cores at 2.2GHz is amazing enough.

[informal]

Slightly OT but from what I see AMD with the MC's has got a excellent product, just need to get the clocks up.
A 12 core MC at 3000mhz would be a force to reckon with.

Yeah that's true.The only problem is they can't crank the clocks that high AND stay in the 105W ACP bracket . What I find interesting is that MC and Lisbon(D1) appear to still not using the low k dielectric tweak AMD implemented in Thuban silicon (E0) which practically made possible for AMD to make a hex core desktop chip with the same clock as QC equivalent(955BE,3.6Ghz @ 3 cores with Turbo) while staying within 125W TDP and actually drawing less than 955BE under full load.If they were to use this major process node tweak on server parts,I think they could bring up the clocks on MC by at least 15% while staying at the same power bands as today.

So you believe that there will be no practical difference between the two? You are very vague.

No,I believe the difference won't be big or impact gameplay that much.Not to mention there won't be 4.8Ghz Deneb chip to actually compare this(unless you do the dry ice session and test it yourself ).But I figure many will be looking to replace their SB or Bulldozer desktop chips with something new in 2014 so testing a Deneb on Dry Ice with the latest GPUs wouldn't be on anyone's priority list .

[SEA]

Since they are working within the same thermal envelope I don't see why it would be meaningless.

If bulldozer performs =12.5% better per core, then it scales to 16 cores just as good as MC scales from 1 to 12 cores..

If bulldozer performs <12.5% better per core, then it scales to 16 cores better than MC scales from 1 to 12 cores.

If bulldozer performs >12.5% better per core, then it scales to 16 cores worse than MC scales from 1 to 12 cores.

1.8 times performance is 90% scaling. 90% of 2 is 1.8.

And we don't know about scaling between modules.

The fact that the scaling within modules isn't the same as the scaling between modules complicates the situation a bit.

Is it just I who think this discussion is going somewhere?
I think that we are a bit closer to conclude that due to modules, bulldozer scale a bit worse per core than MC. And that Bulldozer might have quite high single threaded performance. I think that a module, with all it's shared parts, is better on single threads than a dualcore, but as AMDs states, scales a bit worse. It all comes down to how well BD scales between modules.

OK let me help you to understand.
DB module consists of two quasy-cores, not full cores. They have some shared hardware. You can consider them as hiperthreading. In other words, when one core performance is 100%, 2 cores have only 180%, not 200%.
Same for 8 modules: 800% when used single core within each module, and 1440% using 16 threads with 8 modules and taking ideal case 100% scaling of modules.

It is just very good hiperthreading compared to intel's one.

Regarding scaling of MC and BD you wrote above:
Scaling itself means performance increase when adding same cores. Say, Phenom II 8 cores to 12 cores while having same frequency/IPC per core.. So, again, forget your 12.5. That number is only for 16-threaded Interlagos compared to 12-core MC.
4 module BD will have significantly higher clock speed. Particularly in desktop processors having 125 watt TDP.
And you correctly figured out that single-threaded performance is going to be significantly higher that Phenom II. Actually I see it within 20%-25% considering what we already know.

[tifosi]

Thuban still is to MC what Zambezi is to Interlagos. They have the same relationships. Thuban is AM3 desktop, MC is two such dies for G34. Zambezi is AM3 desktop, Interlagos is two such dies for G34.

Thuban has more in common with Istanbul than with Magny Cours. Magny Cours is an improved new micro architecture which was only to be released for server market, as AMD was hoping that both Fusion and Bulldozer (specifically for desktop) would materialize sooner than this. However, in the server market, it was getting rather too close for comfort with improved Intel server chippery. AMD addressed this situation in the server market with Magny Cours. There is no Magny Cours derived desktop part in the market, but only ones based on Shanghai (Agena) and Istanbul (Thuban).

Which is why i personally have an opinion that we would see a nice jump when it comes to new desktop architecture. However original Phenom was supposed to do so too... So unless things go really that bad for AMD in desktop arena, it is all rosy going forward. Sandy Bridge should have some competition and not just on cost front.

Then again, you have mentioned in another post, Intel still continue with their BS compiler frauds... who knows how much of a negative impact that has on AMD's processor in real world scenario. In one study they found that there is 15-20 odd percent increase in performance, when you change the vendor id from "AMD" to "Genuine Intel". Which is funny, given there's only so much difference between Phenom II and i7's. :P

[Sn0wm@n]

these thread always end up in a never ending argument of cpu X is better then cpu Y .....

[SEA]

No we speak about scaling of hiperthreaded cores...

[Manicdan]

these thread always end up in a never ending argument of cpu X is better then cpu Y .....

are you saying female cpus are better than male cpus?
http://en.wikipedia.org/wiki/X_chromosome

[-Boris-]

No,I believe the difference won't be big or impact gameplay that much.Not to mention there won't be 4.8Ghz Deneb chip to actually compare this(unless you do the dry ice session and test it yourself ).But I figure many will be looking to replace their SB or Bulldozer desktop chips with something new in 2014 so testing a Deneb on Dry Ice with the latest GPUs wouldn't be on anyone's priority list .

Well I say that we don't nedd a 4.8GHz phenom II for that expermient, I say that a chip much faster than Phenom II 940 3GHz will perform much better in games in 2014. And I say that Core i7 is such a chip, and I say Zambezi better be such a chip. Furthermore, it's my opinion that this benchmark is a very bad indicator of real world performance.

I think you are a hypocrite that claims amount of cores and L3 will be important, but at the same time neglect the importance of the architectures properties like IPC and frequency. I say that an i7 that beats a Phenom II with a wide margin in todays games will do so on the games of 2014. And not suddenly become equal just because a GPU limited bench from today showed them to be equal.

OK let me help you to understand.
DB module consists of two quasy-cores, not full cores. They have some shared hardware. You can consider them as hiperthreading. In other words, when one core performance is 100%, 2 cores have only 180%, not 200%.
Same for 8 modules: 800% when used single core within each module, and 1440% using 16 threads with 8 modules and taking ideal case 100% scaling of modules.

It is just very good hiperthreading compared to intel's one.

I know very well about the information that is released on Bulldozer so far. And if we should consider it to be like a good hyperthreading we should call Zambezi a quad core.

Regarding scaling of MC and BD you wrote above:
Scaling itself means performance increase when adding same cores. Say, Phenom II 8 cores to 12 cores while having same frequency/IPC per core.. So, again, forget your 12.5. That number is only for 16-threaded Interlagos compared to 12-core MC.
4 module BD will have significantly higher clock speed. Particularly in desktop processors having 125 watt TDP.
And you correctly figured out that single-threaded performance is going to be significantly higher that Phenom II. Actually I see it within 20%-25% considering what we already know.

I don't think you really understood what I said.
Thuban is a higher clocked singlechip piece for AM3 just as Zambezi is a higher clocked singlechip piece for AM3.

Thuban has more in common with Istanbul than with Magny Cours. Magny Cours is an improved new micro architecture which was only to be released for server market, as AMD was hoping that both Fusion and Bulldozer (specifically for desktop) would materialize sooner than this. However, in the server market, it was getting rather too close for comfort with improved Intel server chippery. AMD addressed this situation in the server market with Magny Cours. There is no Magny Cours derived desktop part in the market, but only ones based on Shanghai (Agena) and Istanbul (Thuban).

Which is why i personally have an opinion that we would see a nice jump when it comes to new desktop architecture. However original Phenom was supposed to do so too... So unless things go really that bad for AMD in desktop arena, it is all rosy going forward. Sandy Bridge should have some competition and not just on cost front.

The differences between Istanbul and Magny Cours is none except that Magny Cours has a new socket. You can see Magny Cours like two Istanbuls in the same package, the new socket makes sure the communicate quickly and each have dual channel DDR3, effectively making it quad channel. The solution with Interlagos will be the same, two Zambezi-like opterons on the same package.

So an Istanbul and a Magny Cours will have the same relationship as a single die bulldozer compared to a dual die bulldozer.


If you still claim that Magny Cours have some improvements over istanbul other than the socket related ones (the same as dual die BD also haves compared to single die BD) please provide a source for you claim.

Then again, you have mentioned in another post, Intel still continue with their BS compiler frauds... who knows how much of a negative impact that has on AMD's processor in real world scenario. In one study they found that there is 15-20 odd percent increase in performance, when you change the vendor id from "AMD" to "Genuine Intel". Which is funny, given there's only so much difference between Phenom II and i7's. :P

Honestly I don't think changing vendor ID is enough. I think the differences in compiling go deeper. Better tailored for intels way of making caches or executing software for instance.

[informal]

Well I say that we don't nedd a 4.8GHz phenom II for that expermient, I say that a chip much faster than Phenom II 940 3GHz will perform much better in games in 2014. And I say that Core i7 is such a chip, and I say Zambezi better be such a chip. Furthermore, it's my opinion that this benchmark is a very bad indicator of real world performance.

I think you are a hypocrite that claims amount of cores and L3 will be important, but at the same time neglect the importance of the architectures properties like IPC and frequency. I say that an i7 that beats a Phenom II with a wide margin in todays games will do so on the games of 2014. And not suddenly become equal just because a GPU limited bench from today showed them to be equal.

Well I think you understand too little when it comes to this subject so I won't bother discussing it with you.

[Calmatory]

I think you are a hypocrite that claims amount of cores and L3 will be important, but at the same time neglect the importance of the architectures properties like IPC and frequency.

I say they are important for AMD. Very important. You talk about games, games and more games. You talk about the ONLY thing which does not scale well with more cores. The thing is that gaming performance doesn't really bring AMD the income it needs. Server platform cost-efficiency brings them the income. So really, I see their approach being the right one.

When it is easier to increase the amount of cores than the IPC, then why not do so? ...especially when the target market scales well with more cores.

[SEA]

I know very well about the information that is released on Bulldozer so far. And if we should consider it to be like a good hyperthreading we should call Zambezi a quad core.

Exactly, and very very good hiperthreading - up to 80% of real core!!!

[-Boris-]

Well I think you understand too little when it comes to this subject so I won't bother discussing it with you.

Well you haven't presented any argument yet. You talk about L3 cache matter and cores matters. But raw speed doesn't matter? It simply doesn't make any sense. There is no reason to why a processor which is capable to greatly outperform another processor in every game of today, suddenly wouldn't do the same in the games of 2014.

[-Boris-]

I say they are important for AMD. Very important. You talk about games, games and more games. You talk about the ONLY thing which does not scale well with more cores. The thing is that gaming performance doesn't really bring AMD the income it needs. Server platform cost-efficiency brings them the income. So really, I see their approach being the right one.

When it is easier to increase the amount of cores than the IPC, then why not do so? ...especially when the target market scales well with more cores.

We aren't talking about what's profitable and what's not. We are talking about things from a consumer point of view. And there is nothing to argue about in multi threaded performance right now. The discussion is about how things are in single threaded performance.

Exactly, and very very good hiperthreading - up to 80% of real core!!!

You miss one little point, hyperthreading can lower performance, this solution might even increase performance in single threaded applications.

[informal]

Well you haven't presented any argument yet. You talk about L3 cache matter and cores matters. But raw speed doesn't matter? It simply doesn't make any sense. There is no reason to why a processor which is capable to greatly outperform another processor in every game of today, suddenly wouldn't do the same in the games of 2014.

"Greatly outperform" in any game(for real ?) at what settings?In real world gaming?Nice try though.

[Manicdan]

@Boris
take a look at steam hardware surveys and see how many gaming rigs have quads
your predictions for gaming are not very valid, by 2014, we might see >50% of games be quad optimized, but not much more than that.

why would i spend 1000$ on a cpu, when 300$ does just as good, and might be 10% worse in 4 years, when 300$ then could get me something that stomps the 1000$ cpu of today

you really should look into the limiting factors for cpus in games and try and assume how that will change in the future. do you really expect the requirements for games to increase 4x, when in the last 4 years have maybe just doubled.

[informal]

Also I see some are describing a BD core as a "half-core".I can't see why that is a case.Is it because of the 10% hit in scaling within a module?

[SEA]

You miss one little point, hyperthreading can lower performance, this solution might even increase performance in single threaded applications.

That is a mistake.
Consider scenario:
1 thread is running on single core of BD delivering a 100% of performance.
2 threads are running on 2 cores within same DB module and delivers 180% of performance.
Now, 2 same threads are running on 2 cores of different DB modules. They deliver 200% of performance. So we saw a performance impact - 180% instead of 200%. And no performance increase...
And this is exactly as it would be with Intel's core with HT.

[SEA]

You can call it hiperthreading, or whatever, but it is a fact: a DB module is not equal to 2 cores and then it should not be considered this way. Just a module, 2-threaded module

[informal]

AMD calls it an optimized dual core,as it is an optimized dual core...
And no it has nothing to do with hyperthreading.