MSI A88XM GAMING: Premium FM2+ Motherboard
Posted: Thu May 22, 2014 11:46 pm
Discussions about Silent Computing
https://www.silentpcreview.com/forums/
https://www.silentpcreview.com/forums/viewtopic.php?f=13&t=67627
Assuming no overclocking, a few of the A10s have higher clocks than the FX-6300 and would thus be slightly faster at tasks that could not use more than 2 threads effectively. Many games do fall into this area, but the gains would be small and probably not noticable. On anything that can use 3 or more threads effectively then the Fx-6300 would be better.dhanson865 wrote:I currently use a FX-6300 on a am3+ MB, what if any FM2+ CPU equals that in performance?
It depends on the application or game. As above, it depends on how many threads can be used effectively. With 2 or less the higher end APUs could meet or slightly beat your FX for a lower TDP.My current CPU is 95W TDP and many of the FM2 CPUs are 65W TDP so if one could equal or beat my FX6300 with a lower TDP that would be another possible plus.
Lots of games are CPU bound and not well threaded. An Nvidia GTX 750 or AMD R7 250 are within your power range and are still fast enough to be CPU limited in some cases.I find it odd to think of AMD cpus as a limiting factor for gaming as I don't buy video cards that use >75W. I can't imagine a HD7750 being CPU limited in many games.
Depends on the game of course, but there are not that many current games that will run well at >1080p with a <75W card, even worse with an APU.I'm currently gaming at 1920x1080 but I'll probably step up in resolution when I get my next monitor and I may do that with the existing CPU and GPU if there isn't a compelling upgrade for either.
washu wrote:Lots of games are CPU bound and not well threaded. An Nvidia GTX 750 or AMD R7 250 are within your power range and are still fast enough to be CPU limited in some cases.
Cistron wrote:I think AMD is still hoping their heterogeneous system architecture (HSA) will take off - but does FM2+ even support those processors?
DirectX 11.x and prior have been counterproductive toward developing highly threaded games. DirectX 12 is supposed to better distribute the CPU tasks. I wish I could find the review where they break it down.quest_for_silence wrote:washu wrote:Lots of games are CPU bound and not well threaded. An Nvidia GTX 750 or AMD R7 250 are within your power range and are still fast enough to be CPU limited in some cases.
Can you kindly point me out something more comprehensive about that?
If you are gaming with the APU as your only GPU and not using a discrete card you aren't going to be CPU bound.washu wrote:Lots of games are CPU bound and not well threaded. An Nvidia GTX 750 or AMD R7 250 are within your power range and are still fast enough to be CPU limited in some cases.Depends on the game of course, but there are not that many current games that will run well at >1080p with a <75W card, even worse with an APU.I'm currently gaming at 1920x1080 but I'll probably step up in resolution when I get my next monitor and I may do that with the existing CPU and GPU if there isn't a compelling upgrade for either.
CA_Steve wrote:Take a look at Blizzard's games. WoW in particular uses 2 cores and is fairly lightweight on the GPU side. Fps is pretty linear with CPU clock and never really tapers off as you speed up the CPU. So, fast CPU with the best single threaded performance is better than throwing more than 2 cores at it. Sort of why a Haswell Pentium can outclass a 4+ core AMD CPU in many game benchmarks.
That thread is confirming what CA_Steve is saying. An i7 (or Xeon close equivalent) is only using 30% CPU. That is about 2 1/2 threads worth. The rest of the 5 1/2 threads the i7 could be using is wasted because WoW can't use them (though in most cases it would help the i7 turbo higher). The only real advantage in WoW that an i7 has over a same-gen Pentium is clock speed.quest_for_silence wrote: Well, but I've also read about people who complain against Core i7 saying it's the limiting factor...
washu wrote:The only real advantage in WoW that an i7 has over a same-gen Pentium is clock speed.
It's precisely the definition of CPU limited. It just doesn't use more than 2 cores.quest_for_silence wrote: Unfortunately, OC'ing set aside, there isn't almost anything really faster than a Core i7 for regular desktops: so I won't call such a setup as "CPU limited", but in case I'd call such a game a somehow crappy piece of software.
washu wrote:It's precisely the definition of CPU limited.
Why would you refuse to believe that? WoW the game engine is 10 years old, but the content is not. They have added many things to it over the years, asking it to do far more than back in 2004. Also, that thread was specifically referencing performance problems in huge raids with many things happening at once.quest_for_silence wrote: even if I don't know WoW at all, I refuse to believe that there isn't a contemporary hardware capable of squeezing substantially more than 25fps (something like 32fps would be pretty the same) from a ten years old game.
washu wrote:It is quite possible (and has been done) to design a map for DOOM that will bring a modern system to its knees.
But exactly this is todays reality: APUs seem to fail for those who game. That in theory they are pretty powerful doesn't matter if real world FPS or subjective gaming performance lacks behind a low cost intel system. AMD was kind of successful in the days when they offered more power at less dollars.quest_for_silence wrote:Well, I wouldn't call that a "design", a "map", a "feature": rather, either a bug, or sort of a computer implementation of the omnipotence paradox.
It's hardly a bug that a app/game attempts to do what is asked of it but runs slowly because the task is demanding. Computer programs should do one of two things: Do what you tell them to, or gracefully fail. Should games just refuse to run if some arbitrary FPS cannot be reached? Just quit with an error if your FPS dropped below 30 in the middle of a heated battle? That is basically what you are saying.quest_for_silence wrote: Well, I wouldn't call that a "design", a "map", a "feature": rather, either a bug, or sort of a computer implementation of the omnipotence paradox.
washu wrote:Do what you tell them to, or gracefully fail.
Assuming that is true with the APU only on the MB is it still true with an APU + dedicated graphics card?Pappnaas wrote:But exactly this is today's reality: APUs seem to fail for those who game.
Yes, it is even more true if you aren't using the GPU part of the APU. The only advantage an AMD APU has over a similarly priced Intel CPU is the GPU part. If you don't use the built in GPU then it comes down to just CPU power and power consumption and Intel beats them soundly on both.dhanson865 wrote: Assuming that is true with the APU only on the MB is it still true with an APU + dedicated graphics card?
All the benchmarks I've seen show #2 to be the case; hybrid graphics don't help enough. For example the best GPU you can use in hybrid mode right now is an R7 250. A Pentium or i3 + R7 250 will give better performance in almost all games than a A10-7850K + R7 250 in hybrid mode and be cheaper.1. GPU still enabled because hybrid graphics are effective at increasing FPS enough for the user to care.
2. GPU disabled because hybrid graphics are a waste of electricity compared to the dedicated GPU.
That is your choice of course, but AMD has lost even their price/performance advantage over Intel in most cases. I bought AMD exclusively, both personally and professionally when it was the better deal. Ever since the Core 2 that has not been the case. Professionally I cannot recommend AMD CPUs anymore as I would likely get fired. In the server space the performance gap is now so great that the higher cost of Intel CPUs is more than offset by the reduced number needed VS AMDs.I may be the rare bird but I have no interest in buying Intel products and will continue to buy alternative CPUs. If AMD can't do it in x86/x64 the market will push towards ARM.
The A10-7850K actually has the "better" core design over the FX-6300 so is slightly faster clock-for clock. I say "better" because it is still a pretty bad design, just refined a bit more. AMD really has no excuse for the Bulldozer/Piledriver/Steamroller core design. They already watched Intel make the same mistake with the P4/Netburst, beat Intel at the time because of it, then went and made the same mistake. Lying about what counts as a core doesn't help any either.So look at the list of processors AMD makes the last time they introduced a new FX CPU at 95W or below was December 2012. The most interesting one on that list for me is
FX-6300 C0 6/3 3.5 GHz 3.8 GHz 4.1 GHz 95 W 2012-10-23
is the A10-7850K the fastest APU they have offered yet?
A10-7850K ?? 4 3.7 GHz N/A 4.0 GHz 95 W 2014-01-14
assuming a game plays nice with 2 cores but doesn't get any gain for the 3rd or higher core (which is common in gaming) and you are using the same discrete graphics in a PCIe x16 slot are these two CPUs roughly on par for gaming or is the difference in core design enough to make the FX CPU still better? If so how much faster does the APU have to be to offset the design difference vs the older FX CPU?
Perhaps somehow questionable: http://www.anandtech.com/bench/product/1200?vs=699 - http://www.pugetsystems.com/labs/articl ... Conclusionwashu wrote:The A10-7850K actually has the "better" core design over the FX-6300 so is slightly faster clock-for clock.
Your first link compares only two multi-threaded tests, so of course the chip with more cores does better. Your second link doesn't even involve the FX-6300, so I'm not sure what point you are trying to make. Neither invalidates what I said.quest_for_silence wrote: Perhaps somehow questionable: http://www.anandtech.com/bench/product/1200?vs=699 - http://www.pugetsystems.com/labs/articl ... Conclusion
At the same time that im intrigued and looking foward to how good will it be, im also a little worried, if its a extremly good overclocker, people could build very cheap setups thinking of popular games that dont require that many cores / threads... like LOL or WOW.washu wrote:The soon to be released "Pentium Anniversary Edition" which can be overclocked should be a great CPU for WoW and other things that arn't threaded well.
This is true, but you can't help those who won't do a bit of research before purchase. Even so, in terms of pure price/performance the new OC Pentium should blow everything else out of the water. Even if a quad+ core would do better at a particular game, it's still going to cost more.Abula wrote:At the same time that im intrigued and looking foward to how good will it be, im also a little worried, if its a extremly good overclocker, people could build very cheap setups thinking of popular games that dont require that many cores / threads... like LOL or WOW.
washu wrote:Your first link compares only two multi-threaded tests, so of course the chip with more cores does better
washu wrote:Your second link doesn't even involve the FX-6300, so I'm not sure what point you are trying to make. Neither invalidates what I said.
APU and FX CPUs both count cores the same but incorrect way. They count 1 module as 2 "cores". The A10-7850K has two modules, so it gets counted as 4 cores. The FX-6300 has 3 modules, and gets counted as 6 cores.quest_for_silence wrote: AFAIK APU and FX don't count "cores" the same way: do they?
This is correct.IIRC the previous APU generation, Richland, sports the same cores of FX-6300:
You didn't read that review very carefully. The A10-7850K looses to the A10-6800K in single core benchmarks simply because of lower clock speed. In pure one thread tasks they are effectively the same at the same clock speed. Where the Kaveri CPU wins is when more than one thread is running on the same module. Because of the shared resources within an AMD module, a second thread will slow down the first. Kaveri reduces, but does not eliminate this slowdown. As the review you linked shows, when multiple threads are being used the 7850K wins over the 6800K even with the clock speed difference because it can scale better. As long as there is more than one thread running (which on a modern OS is always the case) Kaveri will be faster clock for clock. That is the "better" core design. In a lot of cases the higher clock of Richland negates that, which is why I suggested that to dhanson865 if he is looking for a pure 1/2 thread improvement over his FX-6300.so as they said about 7850 "benchmark indicated lower single-core performance", the statement: - "The A10-7850K actually has the "better" core design over the FX-6300 so is slightly faster clock-for clock" - seems somehow questionable.