IPC Test 7: AMD FX Series CPUs! How Does it Compare?

preview_player
Показать описание
We're finally there. Today we test AMD's FX CPU series. Both Bulldozer and Piledriver tested. How does it stack up to Intel's Sandy and Ivy Bridge CPUs?

Previous IPC Tests:

Test Bench Components:

Sources:
Рекомендации по теме
Комментарии
Автор

I am still rocking a Phenom II X6 1100T. Running it @ 4.2GHZ CPU, 2.8GHZ NB. Will be upgrading to Zen 2 next year.

petmey
Автор

💻 My PC is currently a FX-8350, and I'm glad that more games are supporting more cores. It has kept my CPU alive. 👍

TechWeLove
Автор

Very video. Thank's to every one that donated to make this video possible. At the moment - My 990FX / FX-9590 build is still doing well for me. Maybe early next year i may upgrade to a new series.

johnpaulbacon
Автор

AMD was being kept afloat all these years by their GPU deparment. Now it's the other way around.

ToriRocksAmos
Автор

If you ignore intel and just compare amd to amd, it really wasn't that bad. They didn't exactly raise ipc a ton (up some here down some there) but they did increase core count and clocks a ton. . .it just looks much worse because intel happened to make a major ipc leap.

Algorythm
Автор

The answer to how AMD made through that period is simple. AMD CPU's were way cheaper. I bought an FX 9590 with 16mb cache for $240 Canadian dollars back in 2014, while Intel's closest CPU was going for $750 USD.

jasongooden
Автор

I'm under the impression that the FP scheduler can schedule two 128-bit FMAC operations simultaneously.

pyro
Автор

how about adding a Westmere CPU to fill the gap between Lynnfield and SandyBridge?

orthodoxNPC
Автор

No, it is entirely an 8 core CPU. There is no debate on that. FPUs happen to be coprocessors in x86, so saying shared FPUs makes it a quad core is as silly as saying the core count on a i7 depends on how many GPUs you have in SLI.

"Core" has a simple definition. It is(at minimum) something that can fetch instructions and data, execute instructions, and store the result. Not defending the bulldozer design, but it just annoys me that this idea has persisted for so long despite having no basis in computer science/engineering.

amirabudubai
Автор

Unlike a lot of people i'm not in denial that fx8350 had bad ipc. But at least it stayed viable enough paired with a decent gpu to get 1080P 60, even to this day.
While an intel rig with the same set up could get lets say 89 fps, the fx8350 only got 67fps average, the increase did not matter because vsync would cap it at 60fps anyway.
My rig from 6 years ago: Fx8350, radeon 7970, 16gb ddr3 1866mhz ram, 120gb sata ssd (os), 250gb sata ssd (games) can still hit 1080P 60 max in most of the titles I play. I do not play all the latest titles, but just the fact that this machine is still pulling it off is still amazing. This isn't praising AMD, but in reality is actually saying how much the market has slowed down in regards to technical progress in games. E-Sports have also gotten more popular which have also made the need to upgrade to better machines unnecessary if you do not care too much about modern AAA titles like I do. I enjoyed Doom 2016, hated Wolfenstein 2, Prey 2017 was excellent, and Dishonored 2 would have been great if not for the poor optimization.
Out the 4 sample AAAish games I listed, only Dishonored 2 gets less than 35fps, although wolfenstein2 only ran a 45 fps on release and has improved greatly since release past the 60fps mark, but Doom 2016 on the same engine as Wolf2 gets 100fps and looks just as good...
I will tell you one thing though I have recently gotten into the Sims 3 again for the first time in like 8-9 years, and oh boy I do not recommend this game to anyone on an AMD CPU due to its poor optimization / only being coded to work on 2 cores with a little third core offset. (although in 2009 being dual-threaded was a big deal, but add 2 expansions into the mix and even my fx8350 runs like a dog when overclocked to 4.5ghz playing it)

With more modern games using more than 4 cores it is holding the fx8350 on life support just a bit longer so I can wait until the 7nm cpus are released, which is what I have been planning on since the first ryzen release.

pauls
Автор

I got into heated arguments at Amd forums and with even a employee before it came out i knew IPC was going to be lower on Bulldozer as it was common sense looking at the design. Months before launch JF-AMD started to downplay ST performance and i knew it was going to suck. I waited still with my 200$ Sabertooth 990FX board for reviews to come and sure enough tomshardware said the engineering team was trying to "hold the line" on IPC with Bulldozer and increase frequency's to 5ghz yah how dumb is that?


Ended up buying a 8350fx to replace my 1100T that easily OC to 3.9Ghz and my 8350 couldn't go above 4.4 and it was slower in some cases like in emulation which i used often.


So happy Amd got their head together with Ryzen and created a logical architecture.

wii
Автор

test the fx-8xxx and/or fx-9xxx excavator versions?

axxis
Автор

The reason why it does far worse than anyone expected in the benchmarks is because those numbers have been given an inflated level of relevance. I had an FX-8350 for years and never felt the need to change up until last year when I was offered a deal I couldn't refuse on an R7-1700. I'm not saying that the FX was as good as any Intel architecture from Sandy Bridge forward, but I will say that those graphs paint a picture of something that is far beyond human perception. My FX-8350 felt like, well, fast CPU, always. The differences that we're seeing in nanoseconds translate into what would be perceived as identical by the limited human brain. Imagine if the load time of a program on an i7-2600k took two seconds and the FX-8350 took four. Well then, any reviewer with a bit of a flair for showmanship could make a huge chart talking about how the i7-2600k is literally TWICE as fast as the FX-8350 and make a graph that's a ½km long which would make the difference seem huge when it was two seconds that nobody would notice.

I'd had friends over who had their i7-2600k CPUs and they gamed on my rig, playing games like Arkham Asylum, Arckham Knight, Hitman: Evolution and Deus Ex: Human Revolution. They honestly thought that I had overclocked my FX because they expected my PC to be terrible at gaming because that's what reviewers like you, Tom's Hardware, AnandTech, etc. had led them to believe. They were shocked that my "lowly" AMD system was showing FPS numbers that they only dreamed of. Those with the i7-2600k paid $400CAD for their CPUs while those with i5-2500k paid around $350CAD for theirs. They all paid well over $200CAD each for their motherboards.

Meanwhile, I paid just $170CAD for my FX-8350 and $140CAD for my Gigabyte 990FX motherboard (both of which still function perfectly). They also all had GeForce video cards, I believe that the cards they had were GTX 580, GTX 660 and two had the GTX 680. The two that had the GTX 680s were the ones who had i5-2500k CPUs because they had more money left over. Since I had spent so much less on my CPU and motherboard, I was able to pop in twin Gigabyte HD 7970 video cards which meant that my machine, for about the same money, blew theirs completely out of the water because the cost of Intel's CPUs meant that a lot of gaming PCs were badly unbalanced.

I've been building PCs since I was twelve with my first build being a 286-16 after having had an IBM PC model 5150 to play with when my parents upgraded to a 386DX-20. I know that the key to a balanced gaming platform is to be GPU heavy and CPU average. The FX-8350 was more than fast enough to keep out of the way of mt 7970s and not bottleneck them (at least, not in any way that I noticed which is all I care about). The gaming power of twin 7970s was in a completely different level. I could play Arma III at max settings, the 680 could barely play it at medium settings. Gaming-wise, I had more power than the current i7 at the time (I think that it was Devil's Canyon) with a GTX TITAN (according to 3dMark).

The simple truth is that for gaming, the CPU is not unimportant but it pales in comparison to the importance of the GPU. As long as the CPU can stay out of the GPU's way, you're laughing all the way to the bank. I know that I was. ;)

AvroBellow
Автор

Just picked up an AMD FX 8350 for $64, brand new from Newegg.

EscapeVelo
Автор

Bulldozer
A bulldozer is a crawler equipped with a substantial metal plate ... -- Wikipedia
No wonder it is

bivion
Автор

great vid, loving this series. whats next for test #8?

jbres
Автор

hi, will you continue with the serie love it so far!

alcidiow
Автор

Hey! I’m running an fx8350 with an nvidia gtx 1050 which are both considerably overclocked... unfortunately, my motherboard (or maybe cpu) seems to be breaking, and my FPS has decreased by 20 (and ping increased by 30) compared to when I first bought the processor and mobo.
What is a good upgrade to the 8350? Would a Ryzen 5 be a worthy investment?

SquidKing
Автор

Bulldozer/Piledriver was fine for servers, and recently has become better in multi-threaded games! :3

chomper
Автор

dear TGOG, these IPC vids are the best! selecting the tests/benchmarks is always going to be a pain in the butt because some just run better on some architectures, but is it possible some of the particular benchmarks are weighted towards AVX and might be giving sandy bridge an unrealistic advantage? i don't envy your herculean tasks here; it's hard to run a large enough suite of programs to cover everyone's possible workloads lol, there's no way anyone can benchmark dozens of programs without it taking weeks.

zangetsu_the_best_zanpakuto