AMD bulldozer

[Colin.]

hehe frenchy my point it aint as slow as the reviews would lead us belive.

Your point is an interesting one. Indeed lets remember that most the game performance benchmarks were based on older games where the only bottleneck occuring is CPU limitation, this of course reflects badly, perhaps unfairly on the FX architecture. When you look at the more modern games of course there are couple of frames here and there but generally the FX CPU performs up there with the 2600K and 2500K. In fact even the 1100T and the X4 980 perform competitively because the bottleneck here is created by the GPU. Also let us remember that many of the games where a CPU bottleneck occurs are already above 60fps on average even on the FX, so does it really matter at all. I mean +35FPS is safe.

Secondly again lets remember this may well be a perfect server architecture. Multiple cores ready to be fully or half scaled, this is what server software loves and does best.

Finally to address the statement, I have no real problem with the speed of the CPU due to the above reasons however it is the price that effectively forces my hand towards the 2500K. At present much of the software that is needed to run the FX-8150 close to the 2600K does not exist and it just about competes with the 2500K despite being nearly £40-50 more expensive. The FX-8120 is much more competitive and is perhaps even for me worth a look, just buy it and over-clock it, I don't see any reason anyone would buy the 8150, just scale the 8120's frequency up.

 

[Colin.]

Just been reading the benchmarks for the FX-8120, FX-6100 and FX-4100 and the FX-8150 Review at Hexus. Pretty much all come to the same conclusion.

1. AMD is attempting to sell the consumer a server optimised architecture.
2. Do not buy the FX-8150, there is no logical reason to do so. Buy the FX-8120 @£165 and clock to either 3.6Ghz (FX-8150) or 4.6Ghz. Thus I may be swayed to buy a FX-8120 if they arrive here before 1st November.
3. Windows 8 should provide better scheduling and performance should improve yearly according to AMD road-maps.

This reminds me of when Intel first brought Hyper-Threading onto the market, far too early and to damaging reviews. Still I can't wait to see how the Opteron's fare, especially as Cray is building Titan. Probably a good idea as the parallel ability should suit supercomputing.

 

[Phoenix]

Words cannot express my sadness after reading the reviews for something I've waited several years for

 

[Frenchy]

I really dont think theres anything to get "sad" about. They have clearly stated that new OS's will take much greater advantage of it, for example Windows 8 and possibly the newer Linux distributions. I think the architecture itself, which is what you have really been waiting for, is actually quite promising and has great potential, the chip itself may be dissapointing , but generally the first in any line of a new architecture is.

 

[Colin.]

Words cannot express my sadness after reading the reviews for something I've waited several years for

I know how you feel but Frenchy is right, Windows 8 should boost performance and the architecture is promising. My point however remains, would consumers ever utilise more than 4-6 cores? and how strong would each cores need to be?
Bulldozer just screams server architecture, bet the Opteron fanboys will be loving it.

BTW- Does anyone know when PCS will be stocking the FX processors? I may or may not get one but it is worth a look for old time sake.

 

[Frenchy]

I know how you feel but Frenchy is right, Windows 8 should boost performance and the architecture is promising. My point however remains, would consumers ever utilise more than 4-6 cores? and how strong would each cores need to be?
Bulldozer just screams server architecture, bet the Opteron fanboys will be loving it.

BTW- Does anyone know when PCS will be stocking the FX processors? I may or may not get one but it is worth a look for old time sake.

A few years ago people would have asked if people really would ever need dual cores, 2 3GB graphics cards. Very soon general consumers will be using applications that utilize 8 cores, by which time we will develop 16 core architectures and ask the same question again lol.

 

[Colin.]

That is an excellent point. Alternately I just thought that with each generation of quad cores getting a 15-20% increase in speed there might be no need for more cores. Remember they would have to increase the die size for more cores as well. AMD got around this by making weaker cores but how parallel does a CPU need to be, I mean that is why we have GPU's is it not?

Future CPGPU anyone?

 

[Frenchy]

That is an excellent point. Alternately I just thought that with each generation of quad cores getting a 15-20% increase in speed there might be no need for more cores. Remember they would have to increase the die size for more cores as well. AMD got around this by making weaker cores but how parallel does a CPU need to be, I mean that is why we have GPU's is it not?

Future CPGPU anyone?

I dont really know enough about the electronic side of things, but from a processing point of view, parallel threads are extremely important. Obviously with a modern OS multithreading is not an issue and even on a single core, threads can be switched easily, however, there is and always will eb a great advantage to having true concurrency. As applications become more and more pwoerful it would be nice to think that one day each virtual thread for example in the JVM could be allocated its own system thread and its own core.

Again, 20 years ago people would have asked if we really needed more than one core since we can switch threads using the OS and everything gets its allocated time on the cpu, so whats the problem.

 

[Colin.]

Ok, I certainly see the point. I agree actually, it just I thought in the next decade they might use a CPGPU or GPU to do the majority of processing which will by then be parallel but use an old fashioned CPU to do the number crunching that can not be or not as easily parallel processed, this has always been the argument to continue linear processing. It is quite an interesting subject really.

In 20 years time, people will probably make fun of this thread, worried about 4 vs 8 cores while they will be using 256.

 

[Phoenix]

After spending time looking through the reviews there seems to be different results which could be due to the motherboard used. The results shown in this thread make it look pretty good: http://www.tomshardware.co.uk/forum/315805-10-bought-bulldozer-inside and i've found a few reviews that actually give it a great score and show it doing really well like Hardware Heaven: http://www.hardwareheaven.com/revie...sor-vs-core-i7-2600k-review-introduction.html

 

[yurtesen]

After spending time looking through the reviews there seems to be different results which could be due to the motherboard used. The results shown in this thread make it look pretty good: http://www.tomshardware.co.uk/forum/315805-10-bought-bulldozer-inside and i've found a few reviews that actually give it a great score and show it doing really well like Hardware Heaven: http://www.hardwareheaven.com/revie...sor-vs-core-i7-2600k-review-introduction.html

Did you consider most review sites using Intel 2X00[k] processors with overclocked memory controllers (1866mhz is overclocked those processors officially support up to 1333mhz only) compared to standard clocked AMD counterparts? (or then test all processors with 1333mhz memory which will also skew results...)

Nobody wants to see fair benchmark results anyway Things like an OS running several processes simultaneously anyhow is overlooked by excuses like many applications are not multithreaded. Sure, but who runs 1 application at a time? Did those tests running single threads show the real world performance? most of us use DOS right?

Unfortunatley there are so much many factors against AMD's products When the odds are so against AMD, it is even surprising that their products are head to head with Intel in so called benchmark results.

 

[Phoenix]

Did you consider most review sites using Intel 2X00[k] processors with overclocked memory controllers (1866mhz is overclocked those processors officially support up to 1333mhz only) compared to standard clocked AMD counterparts? (or then test all processors with 1333mhz memory which will also skew results...)

Nobody wants to see fair benchmark results anyway Things like an OS running several processes simultaneously anyhow is overlooked by excuses like many applications are not multithreaded. Sure, but who runs 1 application at a time? Did those tests running single threads show the real world performance? most of us use DOS right?

Unfortunatley there are so much many factors against AMD's products When the odds are so against AMD, it is even surprising that their products are head to head with Intel in so called benchmark results.

No, I didn't check the memory speed on the test setup section, most of the tests weren't really showing what normal people use computers for but then again that's not always the point since most people spend a lot of time browsing the internet which CPU's from 4 years ago can do quite well. Something interesting that has been brought up recently is that there's a patch being worked on to fix an issue with the registry in Windows 7 which could result in a 40% improvement in Bulldozers' performance: http://www.overclock.net/t/1143129/quinetiam-amd-fx-bulldozer-registry-fix-40-performance-boost/0_100

Here's a quote from someone explaining why there's a problem in the first place and how the patch aims to fix it:


What is happening is this: The FX 8xxx series of processors has 4 cores, each core with TWO integer pipelines, effectively granting two processor cores for the price of one. (not "hyperthreading" effective, but almost literally additional core effective.)
This is all fine and dandy, except that when Windows (or Linux) goes to issue work to a core in the processor, it picks the core with the least load, and distributes things using a task scheduler. The problem here is that when a single pipeline on a bulldozer core is being utilized both cores are effectively "in use" even though Task Manager sees the cores appropriately, the task scheduler does not. This isn't really anybody's fault, but rather an oversight.
Standard CPU design from a by-the-book standpoint would never add a second integer pipeline to a single core. So nobody ever really thought to code a task scheduler that worked in that situation. So basically, AMD thought outside the box, and Windows got confused.

Oh not to mention that if the scheduler tries to run non-integer work on one of the integer only threads, the CPU returns an error and the command is re-queued and has to sit in line waiting to be executed again. Basically its a scheduler issue, and could definitely cause some HUGE impact in synthetic benchmarks and normal use. And even more issues in a cluttered server environment.

As an eco-friendly person though, the power consumption is still an issue for me :boat: