AMD Gives Bulldozer 6-core a Speed-Bump with FX-6200

[seronx]

Your reasoning is flawed. World of Warcraft does not suddenly become optimised for AMD FX when you run it in Windows 8. The game is the same. As such, the game's being optimised for Intel does not explain the performance gap between Windows 7 with BD and Windows 8 with BD. Furthermore, you do nothing to refute the generally accepted explanation for this phenomenon. That said, if you can provide a valid argument, or better still a credible source boasting one, I'd be very interested.

Your ideology is flawed

Two modules has a peak throughput of sixteen macro-ops much more than the eight macro-ops of one module

When two modules are used you have a higher throughput thus seemingly higher FPS but you then get blockade by the slowest cache

Until then, I think we're better off with the generally accepted explanation, which is that in treating BD CPUs as "normal" 8-cores, Windows 7 isn't (yet) understanding the Bulldozer architecture properly.

Windows 7 sees Bulldozer correctly it has 8 normal cores

Windows 8 will just fix the problem of legacy programs show when a new architecture is introduced that changes the number of decoders

A normal core is registers to ALU clusters
Bulldozer has 8 individual registers to 8 ALU Clusters thus can be called 8 physical cores
Sandy Bridge has 8 registers that have pairs shared to 4 ALU Clusters thus can be called 4 physical cores

Other than that you can go by Database licensing which concludes the amount of cores being the amount of logical cores thus Bulldozer is 8 cores and Sandy Bridge is 8 cores

 

[blibba]

Your ideology is flawed

Lol what? What exactly do you think my ideology is? How is this relevant?

In case you're confused, here's the relevant OED definition of "ideology":

"A systematic scheme of ideas, usually relating to politics, economics, or society and forming the basis of action or policy; a set of beliefs governing conduct."

When two modules are used you have a higher throughput thus seemingly higher FPS but you then get blockade by the slowest cache

So do you get higher FPS, or not?

Windows 7 sees Bulldozer correctly it has 8 normal cores

It has 8 "normal" cores, split into modules of two, with each module sharing a large number of resources, such that two cores from two modules have better performance in many applications than two cores from one module. Agree or disagree?

Windows 8 will just fix the problem of legacy programs show when a new architecture is introduced that changes the number of decoders

Source? If so, we'd see the same performance gain from Windows 8 with Phenom II as we do with Bulldozer (when using legacy applications). Incidentally, I wouldn't describe some of the applications in which BD gets a performance boost in Windows 8 as legacy.

Also, just to zoom out a little here, is your main point that Bulldozer is just a conventional 8 core processor? If so, why does it scale so terribly in some multithreaded tasks? Everyone else's answer is shared resources between cores bottlenecking the modules, and this is backed up by AMD's own accounts of their own architecture.

 

[seronx]

So do you get higher FPS, or not?

You get higher FPS but you won't get stability if you have other modules requiring L3 Cache space, when you have two cores being used in the same module you probably won't have instability as they will only use up to the L2 cache

It has 8 "normal" cores, split into modules of two, with each module sharing a large number of resources, such that two cores from two modules have better performance in many applications than two cores from one module. Agree or disagree?

Disagree

Only 20% of resources are being shared the only problem that can occur is from the code of the program not having alignment to the instruction fetch and instruction decode which will cause problems

2(New apps) vs 3(Legacy apps)
You have bleed with World of Warcraft but instant the bleed is fixed you will see higher FPS with a module vs 2 modules

Source? If so, we'd see the same performance gain from Windows 8 with Phenom II as we do with Bulldozer (when using legacy applications). Incidentally, I wouldn't describe some of the applications in which BD gets a performance boost in Windows 8 as legacy.

Only applications that are legacy will see the boost...

Source still thinks Bulldozer is K10 which has 3 decoders just like Sandy Bridge pretty much
and that is what AMD is using to show of Windows 8 improvements...legacy applications..

 

[blibba]

It doesn't matter what % of resources are being shared if they're causing the holdup.

Furthermore, you respond to very few of my questions in a satisfactory manner - perhaps because English isn't your first language, in which case that isn't something I'd hold against you.

Because of this and because you're so massively out of kilter with what every other expert is saying (notice how I generously offer you an implied compliment there), I'm going to wait for you to back all this up with a credible source.

 

[seronx]

It doesn't matter what % of resources are being shared if they're causing the holdup.

They aren't causing a hold up...they are working properly

Furthermore, you respond to very few of my questions in a satisfactory manner - perhaps because English isn't your first language, in which case that isn't something I'd hold against you.

English is my first language... This forum isn't important so I type fast and pretty much leave after thirty mins and I just finished walking the dog, doing my workout, taking out the trash, and vacuuming my house...so my arms are hurting right now

Because of this and because you're so massively out of kilter with what every other expert is saying (notice how I generously offer you an implied compliment there), I'm going to wait for you to back all this up with a credible source.

http://www.amdzone.com , http://www.semiaccurate.com .

and thanks for the implied compliment

Well first off if you are getting 70 fps @ 2560x1600 .... and Windows 8 only gives you 9 fps you probably won't be able to see it since the monitor is 60Hz....

If you are waiting for Windows 8 you might as well wait for Piledriver which will have up to ~8 IPC uplift and up to ~8 Clockrate uplift(per core)

 

[blibba]

They aren't causing a hold up...they are working properly

http://www.amdzone.com , http://www.semiaccurate.com .

I said credible Besides, you could at least link the actual articles you think disprove the Bulldozer module bottleneck theory.

Well first off if you are getting 70 fps @ 2560x1600 .... and Windows 8 only gives you 9 fps you probably won't be able to see it since the monitor is 60Hz....

This isn't really the point, is it? As explained above, the point is not WoW performance, the point is that one setup is faster than another under certain (common) conditions.

If you are waiting for Windows 8 you might as well wait for Piledriver which will have up to ~8 IPC uplift and up to ~8 Clockrate uplift(per core)

Me? Hell, I'm not in the market for a CPU or an operating system. Even if I was, I'd probably be looking at an i3 - in the absence of a 32nm K10 or equivalent.

 

[seronx]

I said credible Besides, you could at least link the actual articles you think disprove the Bulldozer module bottleneck theory.

You have to look at the forums and look for several dudes they usually post past me and before me...

This isn't really the point, is it? As explained above, the point is not WoW performance, the point is that one setup is faster than another under certain (common) conditions.

Well again completing something in 20 seconds(2M/2C) or completing something in 22 seconds(1M/2C) is relatively unimportant when the guy making the application which has something complete in 20 seconds or 22 seconds to optimize his application for Bulldozer and make the unknown but common workload complete in 18 seconds(1M/2C) is more worth while than saying because CMT is a balanced design we have to go back to CMP which is a over provisioned design and will run slightly better than a balanced design because there is more resources in a over provisioned design.

Me? Hell, I'm not in the market for a CPU or an operating system. Even if I was, I'd probably be looking at an i3 - in the absence of a 32nm K10 or equivalent.

32nm K10 is Llano...and it has a better GPU than the i3....

 

[blibba]

32nm K10 is Llano...and it has a better GPU than the i3....

Aye, but it's not really a good successor to the Phenom II - it's aimed at a different market, and you can see that from the platform, too.

Also, I don't want to pay more than I have to for an integrated GPU I'll never use.

What I really meant is a Phenom II 32nm for the AM3+ platform.

As it happens, I run a C2D and an HD6850 (I won both or I'd have older hardware), and I play all the new games I want to maxed out at 2048*1152, and that's the only demanding thing my PC ever does. So, if I was to be forced to replace it, I'd just want a PC that does all the same as cheaply as possible (idle power consumption being a factor in cost). That probably means picking up an £80 GTX460 or equivalent and an i3, atm. This post is seriously off topic, though

 

[seronx]

What I really meant is a Phenom II 32nm for the AM3+ platform.

Well Phenom II 32nm 8C on the AM3+ will be 2x slower than Sandy Bridge 4C in single thread workloads and 2x slower than Sandy Bridge-E 8C in multithread

Orochi-FX at least gets near the 4C in single thread I think it is more around 1.3-1.5x and multithread is about 1.6-1.75x

---
I forgot to mention this but FX-6100 -> FX-6200 is completely going to bug me out...(one day seronx will check newegg and see 6200 PILEDRIVER OUT NOW?! oh wait it is just that processor...)

 

[blibba]

Given that Phenom II 45nm wasn't even outperformed by 100% by SB, I don't see how you can say that Phenom II 32nm would be (I'm assuming that the jump to 32nm would yield higher clock speeds or at least better power efficiency). Current FX processors are even further behind SB in single threaded workloads (and some multi-threaded workloads) than the Phenom X6 was.

http://www.anandtech.com/bench/Product/362?vs=288

http://www.anandtech.com/bench/Product/203?vs=434

The first link shows the top Phenom II 4C being outperformed by less than 100% by the top SB 4C/4T.

The second link shows the top Phenom II 6C outperforming Bulldozer in a few single threaded tasks and very much holding its own in many multithreaded tasks. Take a look at single threaded Cinebench, for example.

 

[seronx]

Given that Phenom II 45nm wasn't even outperformed by 100% by SB, I don't see how you can say that Phenom II 32nm would be (I'm assuming that the jump to 32nm would yield higher clock speeds or at least better power efficiency). Current FX processors are even further behind SB in single threaded workloads (and some multi-threaded workloads) than the Phenom X6 was.

http://www.anandtech.com/bench/Product/362?vs=288

http://www.anandtech.com/bench/Product/203?vs=434

The first link shows the top Phenom II 4C being outperformed by less than 100% by the top SB 4C/4T.

The second link shows the top Phenom II 6C outperforming Bulldozer in a few single threaded tasks and very much holding its own in many multithreaded tasks. Take a look at single threaded Cinebench, for example.

http://www.anandtech.com/bench/Product/88?vs=363

It isn't always 2x but majority of the time you will be near 2x

Fallout 3 near 2x
WinRAR near 2x
Par2 near 2x
Blender near 2x
FLV Creation near 2x
R10 ST/MT near 2x
Adobe CS4 near 2x

Other than a lot of programs don't support FMA4/AVX/XOP "-march:bdver1" thus you aren't really seeing full IPC with Bulldozer while alot of programs show support for AVX for SB "-QxAVX"
While plenty of programs show support for /archSSE3 and /marchSSE3

Full SB, Full K10, Half BD isn't really fair game

 

[etayorius]

I wont upgrade my PhenomII 810 to ShitDoZer untill AMD hits 4Ghz without Turbo and TDP lowers just a bit.

 

[blibba]

http://www.anandtech.com/bench/Product/88?vs=363

It isn't always 2x but majority of the time you will be near 2x

Fallout 3 near 2x
WinRAR near 2x
Par2 near 2x
Blender near 2x
FLV Creation near 2x
R10 ST/MT near 2x
Adobe CS4 near 2x

In most of those, if you look at the numbers, it's closer to 1.5x. Huge difference between that and 2x. And those are the most extreme examples. Actually, some of what you list is closer to 1.3x.

Other than a lot of programs don't support FMA4/AVX/XOP "-march:bdver1" thus you aren't really seeing full IPC with Bulldozer while alot of programs show support for AVX for SB "-QxAVX"
While plenty of programs show support for /archSSE3 and /marchSSE3

Full SB, Full K10, Half BD isn't really fair game

Well it is fair game if those are the applications that you're going to use. If my CPU is slow because software doesn't take full advantage of it, my CPU is still slow.

 

[JustaTinkerer]

I am honesty sick BD bashing....though i keep reading, anyone running windows 8 full time?
No....oh why is that i hear the folks calling....!
Cos its not retail and it doesnt fix s**t, BD is what it is right now live with it

I picked up my FX-8120 for £125 on sale here, no longer on sale (look hard enough) because of the reviews and bad press no doubt.
Anyone able to tell me if I can get an Intel for the same cash that does the same job.
I recode, game, compress, jeez a bit of everything?

I don't care about AMD vs Intel (last processor was a Intel) all anyone who call's themselves an overclocker should care about is price vs performance.

FX owners that adopted early I do feel for.

Windows 8 is not the FX fix, its just a pipe dream, if you think windows 8 will fix everything like a fairy godmother then you are mistaken.
By the time it comes out retail Intel will have stepped the game up 2 levels....is FX going to "come in to its own" .....we can only wish...well cinders did get to the ball i suppose

 

[blibba]

Windows 8 is not the FX fix, its just a pipe dream, if you think windows 8 will fix everything like a fairy godmother then you are mistaken.

Afaik nobody thinks that.

Also, grats on the bargain. Asking others to match that is a little harsh, though.

There are Phenom II x6s and socket 1366 i7s in that price range very occasionally, both of which are comparable on performance.

 

[seronx]

In most of those, if you look at the numbers, it's closer to 1.5x. Huge difference between that and 2x. And those are the most extreme examples. Actually, some of what you list is closer to 1.3x.

You have input the Intel CPU has a longer pipeline and wins most of the time... pushing it even more towards 2x

Well it is fair game if those are the applications that you're going to use. If my CPU is slow because software doesn't take full advantage of it, my CPU is still slow.

Most applications don't care about ISAs but most benchmarks do

In most cases in the native environment some of these applications won't exist
(x87, MMX, 3dnow!, SSE(64bit) can't exist in x86-64, in x86-64 you have to use SSE2,SSE3,SSE4,SSE5(AVX+FMA+AVX2+Gather+XOP)

In most music conversion you see MMX and SSE being most used...while in 64bit applications of music conversions you see SSE4 being used the 64bit music converter is faster than the 32bit music converter but the 32bit version is still being more used...

Consumers = Relatively Stupid....in these cases

Smart Consumers like myself know to wait for applications to use the new ISAs before jumping boat or listening to non-important reviewers trying to persuade unsmart consumers in making dumb decisions

 

[Mussels]

What you need to consider is that everything 96W and up must be marked with the 125W TDP envelope.

thank you for pointing that out. people always think these wattage numbers are set in stone, when really its just what envelope they fit into.

 

[Thefumigator]

What you need to consider is that everything 96W and up must be marked with the 125W TDP envelope.
So we really don't know how much more power does it use with a ~15% clock increase.
So until a full review is up we won't know if there are any improvements.

Except llano, 100W

I am honesty sick BD bashing....though i keep reading, anyone running windows 8 full time?
No....oh why is that i hear the folks calling....!
Cos its not retail and it doesnt fix s**t, BD is what it is right now live with it

I picked up my FX-8120 for £125 on sale here, no longer on sale (look hard enough) because of the reviews and bad press no doubt.
Anyone able to tell me if I can get an Intel for the same cash that does the same job.
I recode, game, compress, jeez a bit of everything?

The thing isn't bad, it just sells well anyway. I'm also sick of BD bashing, but hey, you have to be honest, we expected more. I actually expected to be comparable with i7, but hey its not far. Also guess what? You could get a G34 board and put a 16 core bulldozer or 12 core bulldozer, yeah, the ones appeared in newegg, 6 modules way cheaper than the 8 modules one, I'm planning building one for myself to make a heavy multitasker on the cheap. Also if the board is dual G34 you don't have to fill both sockets to make the thing run, and this is great.

 

[Mussels]

Except llano, 100W

the power envelopes are set per socket. llano is FM1, bulldozer is AM3+


AM3+ follows the same wattage envelops as AM2/2+/3, which is 65, 95, 125.

if its 96W or above, it has to be specified as 125W.

 

[[H]@RD5TUFF]

seems rather pointless

 

[entropy13]

seems rather pointless

Only in the sense of comparing them with other processor series, whether from AMD or Intel. It populates the Bulldozer line-up quite well however. It's when you go beyond them that things doesn't go "well", to put it mildly.

 

[nt300]

Are we finaly done with the pissing contest. Now I should get myself a 8 core Dozer.

 

[DannibusX]

 

[ViperXTR]

and i heard intel is outing the i5 2550K this christmas? >_>

 

[Zubasa]

I wont upgrade my PhenomII 810 to ShitDoZer untill AMD hits 4Ghz without Turbo and TDP lowers just a bit.

The FX-4100 can easily go pass 4Ghz already, it is actually a decent upgrade for the PII 810 aka C2 Phenom IIs.
Not to mention memory performance is a crap load better. The IMC on the old 810 is bottle-necking even DDR3 1600.