AMD FX-8350 - "Piledriver" for AMD Socket AM3+

[TheoneandonlyMrK]

It was very interesting to see everyone guessing at stuff prior to hte launch, to be honest.

yeh i had thought clock mesh was in, apparently not:shadedshu, still good review, I was eagerly looking forward to buying an 8350 but im homeless now come weekend so that plans back on the back burner , I must say though in all the threads where Amd bashing goes on ive been sat with a 960T ,apparently some oll tat that cant do much these days and only 50gops of omph and its still ultras the world at 1080p hearing people call many a better chip rotten ,fussy odd world isnt it

cheers for the read dave , moar pages next time please ,it ran out too quick as ever

Also notice that I listed my Intel i7-3820 out of my gaming rig for sale. Guess what will replace it?

, a recomendation indeed..

 

[cadaveca]

3970X?

I no has Intel contact. I'd love one of those though, but alas...I still buy my Intel chips @ retail.


harr harr.

yeh i had thought clock mesh was in, apparently not:shadedshu, still good review, I was eagerly looking forward to buying an 8350 but im homeless now come weekend so that plans back on the back burner , I must say though in all the threads where Amd bashing goes on ive been sat with a 960T ,apparently some oll tat that cant do much these days and only 50gops of omph and its still ultras the world at 1080p hearing people call many a better chip rotten ,fussy odd world isnt it

cheers for the read dave , moar pages next time please ,it ran out too quick as ever




, a recomendation indeed..

Unfortunately there was limited time for testing due to waiting for BIOS, and then having to retest everything with the new BIOS, waitiing for drivers...etc...AMD did a good job of keeping things quiet before the launch. I asked every single OEM for a board to test with...not one answered. And AMD didn't send one. Oh well, thankfully i kept one on-hand..I just recently cleaned out a tonne of boards since the wife was starting to eye them up a bit too much...

 

[cdawall]

Still way too slow single thread performance. And very uncertain performer in general. Sometimes very fast, sometimes very slow. Sure, it is better than old Bulldozer but I would find it hard to recommend this over 3570K for general use ie. gaming.

Why? Because some poorly coded games at a 10 millisecond difference? If we are going off of that all of the Phenom II owners better just sit were they are. Single IPC is dead welcome to 2010. I could honestly care less if a Pentium MMX chip beat it in single IPC that is not what these chips are designed to do and the sheer fact that they are even keeping up is astounding.

When we go brake into actual multithreading

Some how AMD is beating the $350 processor and competing with the $999 one.

Even in cinebench which is immensely Intel biased.

AMD beats all of the chips in its price class.

Hmmm yet again

And WTF is this? something optimized for all different CPU's out there (not just Intel) and the AMD core kicks ass. Likely due to AVX instructions, but still.

Very few things are single-threaded at this point, and even Intel is pushing multi-threading to devs.

Weird that you mean for the past 5-6 years everyone has had dual cores or better?

review snip

I noticed in your system builds the Intel boxes has 2133mhz ram in one and 2666mhz ram in the other while the AMD had 1866mhz ram what was the actual clock on all of the ram kits?

 

[cadaveca]

Why? Because some poorly coded games at a 10 millisecond difference? If we are going off of that all of the Phenom II owners better just sit were they are. Single IPC is dead welcome to 2010. I could honestly care less if a Pentium MMX chip beat it in single IPC that is not what these chips are designed to do and the sheer fact that they are even keeping up is astounding.

When we go brake into actual multithreading

http://img.techpowerup.org/121023/Capture040602.jpg

Some how AMD is beating the $350 processor and competing with the $999 one.

Even in cinebench which is immensely Intel biased.

http://img.techpowerup.org/121023/Capture041.jpg

AMD beats all of the chips in its price class.

http://img.techpowerup.org/121023/Capture042.jpg

Hmmm yet again

http://img.techpowerup.org/121023/Capture043.jpg

And WTF is this? something optimized for all different CPU's out there (not just Intel) and the AMD core kicks ass. Likely due to AVX instructions, but still.



Weird that you mean for the past 5-6 years everyone has had dual cores or better?



I noticed in your system builds the Intel boxes has 2133mhz ram in one and 2666mhz ram in the other while the AMD had 1866mhz ram what was the actual clock on all of the ram kits?

1600 MHz for the 3770K, 1600 Mhz for the 3570K. board BIOSes made XMP profiles not work, unfortunately, and i thought since 1866 Mhz is what FX-8350 supports default, 1600 Mhz for IVB default, so the speeds were perfect.

3770K was 11-11-11-28, 3570K was 9-9-9-24.

 

[cdawall]

1600 MHz for the 3770K, 1600 Mhz for the 3570K. board BIOSes made XMP profiles not work, unfortunately, and i thought since 1866 Mhz is what FX-8350 supports default, 1600 Mhz for IVB default, so the speeds were perfect.

Thanks that what I was curious about

My mind is still blown from that zlib benchmark I cannot wait until cinebench is recoded to allow AMD to use AVX.

 

[cadaveca]

Thanks that what I was curious about

My mind is still blown from that zlib benchmark I cannot wait until cinebench is recoded to allow AMD to use AVX.

i think it has been, but hhe build was not released in time for testing...i think AMD gave me a build on Friday?

you can see in the power consumption when doing certain tasks that hte software isn't using all of the chip it can, in some instances.

 

[tehehe]

Still bad for games. If multithreading in gaming is the future then I will buy adequate cpu when this future comes, not now. These cpus are simply ahead of their time and that's not a compliment.

 

[cadaveca]

Still bad for games.

Compared to what?

I cannot say it's bad for gaming. Been playing a fair bit of BF3 and some Rocksmith as of late though, and not much else.

 

[BlackOmega]

Suripse . . .. err wait nope still the same garbage as the last round of AMD chips, when will AMD get their act together and make a real chip.:shadedshu

Same garbage? Hardly.

There's been so much improvement over BD it's fairly striking. 1866MHz memory stock! That's huge in and of itself. Need I remind you that AMD chips net huge performance gains from ocing the memory.
Performance is up, power consumption is down, and for a lot of 990 users it's a direct drop in.

Is it going to compete with Intel? Only if it's priced accordingly. Performance wise, AMD simply can't compete. Intel R&D spends more in a month than AMD's YEARLY budget. So no surprise that Intel chips are faster. Actually, it's rather impressive that a relatively small company like AMD can compete at all.


Regarding humansmokes comment saying that using anything less than LinX to stress test is "cheating", I find it rather funny. LinX/IBT are good for testing Intel silicon, not AMD. From my own and some friends testing AMD overclocks, those two tests totally suck at testing AMD's and have been removed from my stress testing regimen.
A friend ran 18 hours of LinX on his AMD (Max everything), passed with flying colors. Turned on F@H and it bsod'd within 5 minutes.
Prime95 and F@H are still the best stress tests for AMD's. Too bad that a little known stress tester, S&M, didn't continue to the windows 7 era. It would, by far, get my AMD silicon the hottest.

 

[cdawall]

i think it has been, but hhe build was not released in time for testing...i think AMD gave me a build on Friday?

you can see in the power consumption when doing certain tasks that hte software isn't using all of the chip it can, in some instances.

Are you planning an update with games, new benchmarks etc?

Still bad for games. If multithreading in gaming is the future then I will buy adequate cpu when this future comes, not now. These cpus are simply ahead of their time and that's not a compliment.

Well for the $100-200 you save over a 3770K which is apparently the best processor on Earth according the team Intel you can take the 7870 you budgeted for and buy a 7970. Good news about that instead of .5-5FPS of difference now you are getting 15-20FPS of difference. So which is the better value for games in your book? A 3770K with a midrange card or 8350 with a high end card? I sure plan on the high end card.

 

[cadaveca]

Are you planning an update with games, new benchmarks etc?

Provide me a list of games with automated benchamrks, and I will gladly run them, provided I own the title.. Play-through benchmarks are not an option.


I jsut purchased SAleeping Dogs on the Steam sale, so that might be added soon..I need to check and see if these games show differences first, etc...I do tend to vet my tests to make sure they actaully show true perforamcne differences, or if they show garbage.

Like Sniper Elite V2... I bench for every motherboard..shows ZERO differences, but for CPU testing, with big speed differences, it does. So it's in my CPU/APU reviews, but not board or memory reviews. It might make hte cut for memory...testing is still underway.

Heck, I'm re-testing and APU boasrd right this second because of the driver updates. SB driver improves SATA numbers for FM2, apparantly.

 

[STCNE]

Provide me a list of games with automated benchamrks, and I will gladly run them, provided I own the title.. Play-through benchmarks are not an option.


I jsut purchased SAleeping Dogs on the Steam sale, so that might be added soon..I need to check and see if these games show differences first, etc...I do tend to vet my tests to make sure they actaully show true perforamcne differences, or if they show garbage.

Like Sniper Elite V2... I bench for every motherboard..shows ZERO differences, but for CPU testing, with big speed differences, it does. So it's in my CPU/APU reviews, but not board or memory reviews. It might make hte cut for memory...testing is still underway.

Heck, I'm re-testing and APU boasrd right this second because of the driver updates. SB driver improves SATA numbers for FM2, apparantly.

GTAIV and ARMA 2 have built in benchmarks like that and tend to be CPU heavy games. Would be interesting to see how the chip stands up. GTAIV needed a pretty beefy CPU last I played(over a year ago). Ran decently on my old i5 750, was only smooth after I did some OCing.

 

[cdawall]

Provide me a list of games with automated benchamrks, and I will gladly run them, provided I own the title.. Play-through benchmarks are not an option.


I jsut purchased SAleeping Dogs on the Steam sale, so that might be added soon..I need to check and see if these games show differences first, etc...I do tend to vet my tests to make sure they actaully show true perforamcne differences, or if they show garbage.

Like Sniper Elite V2... I bench for every motherboard..shows ZERO differences, but for CPU testing, with big speed differences, it does. So it's in my CPU/APU reviews, but not board or memory reviews. It might make hte cut for memory...testing is still underway.

Heck, I'm re-testing and APU boasrd right this second because of the driver updates. SB driver improves SATA numbers for FM2, apparantly.

Crysis, dirt showdown have ones. I will have to look around for what else does.

 

[cadaveca]

Crysis, dirt showdown have ones. I will have to look around for what else does.

Yeah, Perhaps start a thread? We can poll other users, etc..I'm always open to changing testing, this and hte APU review are my first CPU-type reviews, so I am definitely open to changing things a bit.


I jsut can't devote too much time to testing though, and each test MUST show an actual difference. If there is less than 1 FPS difference, I won't use it. If the results are not repeatable, I won't use it. If I cannot affect results by tweaking system speeds, I won't use it.


I'm kind of a benchmarking snob, these days. However, I've been so busy with reviews teh past year or so that I am not really up to date on what's out there game-wise.

 

[cdawall]

Yeah, Perhaps start a thread? We can poll other users, etc..I'm always open to changing testing, this and hte APU review are my first CPU-type reviews, so I am definitely open to changing things a bit.


I jsut can't devote too much time to testing though, and each test MUST show an actual difference. If there is less than 1 FPS difference, I won't use it. If the results are not repeatable, I won't use it. If I cannot affect results by tweaking system speeds, I won't use it.


I'm kind of a benchmarking snob, these days. However, I've been so busy with reviews teh past year or so that I am not really up to date on what's out there game-wise.

Arma II requires some very high end SSD's to really show differences. Crysis is the easiest one and shows tons of difference. Same goes for Dirt will post a couple quickies between stock and 3.9ghz on my chip. In a new thread obviously.

http://www.techpowerup.com/forums/showthread.php?p=2755212#post2755212

^Thread for discussion^

 

[EarthDog]

Compared to what?

I cannot say it's bad for gaming. Been playing a fair bit of BF3 and some Rocksmith as of late though, and not much else.

Have you had a chance to compare BF3 with the 8350 and 3570K/3770K at the same clock and memory speeds? That should isolate things as best possible to show CPU differences in gaming. I dont think there will be much difference with midrange cards, but when you jump up and especially in SLI/Crossfire, I have a hunch that The Intel will still pull away as it has in the past.

This is in OLD LINK, but, a reference point none the less. Im sure it will depend on the title, resolution and cards used, but... it was true two years ago with less powerful CPU's... I dont imagine it to change much now? No clue... but it would be interesting to see actual H2H competition as best we can (same memory capacity/speed/timings, same CPU clockspeeds, and of course the same GPU).

 

[cadaveca]

Have you had a chance to compare BF3 with the 8350 and 3570K/3770K at the same clock and memory speeds? That should isolate things as best possible to show CPU differences in gaming. I dont think there will be much difference with midrange cards, but when you jump up and especially in SLI/Crossfire, I have a hunch that The Intel will still pull away as it has in the past.

3570k and 8350 are neck-and-neck with 7950 Crossfire at stock, which is what I used for testing.


I have not done a "clock-for-clock" compare between Intel/AMD, as i see no point. Then voltages might be differnt, and then that could be comapred...I'm not into comparing how every aspect of the chips are different. I am more interested in showing them as they will be used, rather than hypothetical situations. I mean with hypotheicals, I might as well post nothing but Futuremark benchmarks then.

 

[suraswami]

Provide me a list of games with automated benchamrks, and I will gladly run them, provided I own the title.. Play-through benchmarks are not an option.


I jsut purchased SAleeping Dogs on the Steam sale, so that might be added soon..I need to check and see if these games show differences first, etc...I do tend to vet my tests to make sure they actaully show true perforamcne differences, or if they show garbage.

Like Sniper Elite V2... I bench for every motherboard..shows ZERO differences, but for CPU testing, with big speed differences, it does. So it's in my CPU/APU reviews, but not board or memory reviews. It might make hte cut for memory...testing is still underway.

Heck, I'm re-testing and APU boasrd right this second because of the driver updates. SB driver improves SATA numbers for FM2, apparantly.

Nice review Dave.

People like me use these desktop parts for home brewed servers, VMs and databases for study and practice. Any chance you can test the 8350 for SQL Server 2012 multi-thread performance and VMware multi vms performance?

 

[EarthDog]

Wait, arent they usually BOTH used overclocked, even here at TPU? Isnt Crossfire/SLI gaming a 'real' situation? I could care less about the synthetics...

I mean I expect having a 600Mhz clockspeed difference you wont see much difference, but what about both of these at 4.6Ghz? A common overclock for the 3570K and likely the 8350? What do games show across CPU limited reso's (say 1280x1024) and GPU limited (say 1920x1200)? In the words of Led Zepplin, does 'the song remain the same'?

 

[tehehe]

Compared to what?

I cannot say it's bad for gaming. Been playing a fair bit of BF3 and some Rocksmith as of late though, and not much else.

Compared to identically priced 2500k. If a game is cpu intensive then Intel wins. Otherwise it's a tie. So Intel is better for people that game a lot - like me. AMD is bad in this competition as far as games are concerned.

 

[cadaveca]

Compared to identically priced 2500k. If a game is cpu intensive then Intel wins. Otherwise it's a tie. So Intel is better for people that game a lot - like me. AMD is bad in this competition as far as games are concerned.

I do not really know of many games that are really CPU-limited though. Like maybe 5...

And then we are tlaknig about FPS that do not make a real difference to actual game playability.


Numbers are one thing, but actual experience offered is another.

 

[Maban]

How does it do at 5GHz with F@H?

 

[cdawall]

Wait, arent they usually BOTH used overclocked, even here at TPU? Isnt Crossfire/SLI gaming a 'real' situation? I could care less about the synthetics...

A lot of people don't overclock as much as you think. Especially in crossfire and SLi when the temps really start to climb.

I mean I expect having a 600Mhz clockspeed difference you wont see much difference, but what about both of these at 4.6Ghz? A common overclock for the 3570K and likely the 8350? What do games show across CPU limited reso's (say 1280x1024) and GPU limited (say 1920x1200)? In the words of Led Zepplin, does 'the song remain the same'?

The 8350 in every single benchmark review posted has done 5ghz stable. Why do you think they are only going to get 4.6ghz now? Is there some Intel conspiracy in your pocket that is going to make them unstable if they start performing better? I expect to see the actual later retail model chips in the 5.5ghz range without problems. Dave got his up to 5ghz on the AMD AIO watercooling which is no better than an H70. That is about the same cooling power as a high end aircooler. Real water cooling will only do better.

 

[Dent1]

It comes down to balance to.

Compare similarly priced Intel and AMD solutions, Intel wins majority of games, but loses in a landslide in most multi threaded work related tasks. Looking at Guru of 3D amongst a few other review sites AMD solution wins majority of multi threaded tasks which is a huge positive. IMO the AMD solution is a clear winner for anyone using their PC for work more than 40% of the time.

Compared to identically priced 2500k. If a game is cpu intensive then Intel wins. Otherwise it's a tie. So Intel is better for people that game a lot - like me. AMD is bad in this competition as far as games are concerned.

In your case, a lot the gaming benchmarks were single threaded or old games. For somebody holding onto their rig for upto 5 years AMD might still be the optimal solution for gaming too as with every new release we're seeing more emphasis on multi core compatibility. In the next couple of years if games utilised cores like Cinebench to dominate Intel, Piledrivers could stand a great chance at dominating gaming benches too with maturity.

In closing, 2500k might dominate gaming today, but might lose by a greater margin to Piledriver tomorrow - Think about it.

 

[EarthDog]

LOL, it was just a random number cdawall, no Intel conspiracy! Fine, jack up the 3570K to 4.9Ghz or 5Ghz like I occasionally run mine at (H100). The point was it has a 600Mhz headstart so I would expect things to be a lot closer than if both with clocked the same... regardless of that clock speed.

Though I would doubt retail is going to hit 5.5Ghz 24/7 stable with 'normal' cooling (ambient water or less). I could be wrong, and actually hope I am... until then, its all speculation.