AMD FX 8320's performance in games. Is it that bad?

[Gikaseixas]

I suppose you could try Giga ud5 or better

UD5 is great, i have had no problems at all
UD7 is even beefier, they went all out

 

[MiddleOfTheRoad]

Not even remotely close to a 2500K. A 2500K can be 50% faster in games than a 8320.

Totally untrue. An FX-8350 can play most games at levels similar to a 2500k -- an FX 8350 can run some multi-threaded apps as fast (and occassionally faster) than a 3770K. It really depends how much the game relies on Hyperthreading -- games with heavy HT are the ones that pull substantially ahead. A lot of modern games are bottle-necked by the GPU.... In those instances, the CPU really doesn't even matter much.

 

[Gikaseixas]

If you look at the shogun 2 results in Shintai's first link, the 8350 is getting less than 30 fps ave and less than 20 minimum, and the 8320 is clocked 10% lower still, so I think a CPU bottleneck at least in that game is entirely plausible.

Dude, i guarantee you that if you use one, your perception will change. There's nothing a slight oc doesn't cure

 

[AtenRa]

If you look at the shogun 2 results in Shintai's first link, the 8350 is getting less than 30 fps ave and less than 20 minimum, and the 8320 is clocked 10% lower still, so I think a CPU bottleneck at least in that game is entirely plausible.

Like Civ V build in benchmark, Total War CPU benchmark is not how you play the game. You don’t zoom so close to actual combat when you play a battle, because you always need to the see enemy army movements and the rest of your army so to counter and coordinate its movements.
In both games most of the time you spend in half way zoom, not fool zoom which doesn’t make it completely CPU bound.

 

[MiddleOfTheRoad]

Folk on the internet always say that AMD cpus are not good , then post some benchmarks. I actually believed it and went from an FX8350 to i7 3770K. Did not notice a huge increase in frames ... in fact it was single digits

Anyway I recently built an i5 pc for a friend and compared it against the FX8350

Thief on i5 was 51.1 (highest settings) and on the FX it was 43.9. That was the biggest difference

Tomb Raider was 63.9 on the i5 and 60 on the FX

Rubbish gaming CPU ? I don't think so

I've seen similar results. You really have to spend probably about $350+ on an Intel processor to get a substantial boost over an FX-8320/8350.

 

[SPBHM]

Folk on the internet always say that AMD cpus are not good , then post some benchmarks. I actually believed it and went from an FX8350 to i7 3770K. Did not notice a huge increase in frames ... in fact it was single digits

Anyway I recently built an i5 pc for a friend and compared it against the FX8350

Thief on i5 was 51.1 (highest settings) and on the FX it was 43.9. That was the biggest difference

Tomb Raider was 63.9 on the i5 and 60 on the FX

Rubbish gaming CPU ? I don't think so

are these numbers from the built in benchmark?

the built in benchmark on TR is kind of useless for CPU comparison, I would be curious to see a proper comparison on the most cpu dependent areas like the fight at the shanty town, or some bad parts of mountain village (with the highest level of detail settings)

anyway, if you have a limited number of games and requirements and a GPU limited situation it wouldn't be a surprise to see slower CPUs performing the same.

the 2500K can be a lot faster than the 8320 for many games or not, and close for others, it depends on the games you play and how you play...

 

[sm625]

Wow. It is hard to believe that a shared FPU in itself can possibly cause delays of that magnitude. To go from 500 on an i5 to over 12000 on the AMD tells me that there has to be some serious cache/memory thrashing or something on that order of magnitude. This is exactly the kind of obvious bug that could be profiled out, if AMD had any money to spend.

 

[blackened23]

Someone may find them usable in GPU limited games but not every game is GPU limited, and the big problem with the FX8320 or 50 is that they are inconsistent. You can find games where they are easily outperformed by an i3. Then you find another game where performance is acceptable. And those games where performance is acceptable, you're usually looking at performance of 3 generation old SB level performance. The case where they perform well are very few and far between, whereas the intel platform will give you consistency no matter what and the cost isn't that much greater really.

For the tradeoffs of the platform and relative in-expensiveness of intel's i5 CPUs, I don't think they're worth it really. Unless you already have a motherboard in place and can get the FX chip itself for cheap. If you want an overclocking FX CPU motherboard, you need a good one, and those aren't really a bargain. And you still have the issue of wildly inconsistent performance....an i5 Haswell is just better all around IMO, sorry to say if you want the best gaming experience that's going to be intel.

If you want a bargain type of chip AMD has a few good ones, but if you're talking the 8320 or 8350 with an overclocking motherboard, huge CPU cooler, eh. Not worth it.

 

[BSim500]

Someone may find them usable in GPU limited games but not every game is GPU limited, and the big problem with the FX8320 or 50 is that they are inconsistent. You can find games where they are easily outperformed by an i3. Then you find another game where performance is acceptable. And those games where performance is acceptable, you're usually looking at performance of 3 generation old SB level performance. The case where they perform well are very few and far between, whereas the intel platform will give you consistency no matter what and the cost isn't that much greater really.

Agreed. I picked up my i5-3570 (non K) for $10 less than the average FX-8350 was listed (Newegg) at the time, so the usual "Intel's cost SO much more" fanboy claims went straight out the window. Even at stock 3.4GHz it regularly scored 5-30% higher fps overall in dozens out of the +200 odd games I own including +20% gains on Skyrim, Oblivion, Starcraft 2, Thief, etc. With a mild OC to 4GHz (same clock), perf was up to 45% higher whilst drawing less than half the power (measured).

The real issue is more min fps than avg fps. Everything just felt much smoother when the "worst" fps in the "busiest" areas was 40% higher on the Intel in quite a few games. Same with up to 10x worse frame latency. Even some 10-15 year old games will still totally saturate an FX resulting in 35-40fps FX vs 50-70fps on an i5 due to their over-reliance on demanding everything be theoretically perfectly threaded as a substitute for improving IPC. If you have an existing AMD board, they make sense as a "drop in", but if you're doing a full upgrade and want consistent high performance, they're not very appealing if you have a large game collection that consists of more than just Crysis 3 & Watch Dogs...

 

[bononos]

Folk on the internet always say that AMD cpus are not good , then post some benchmarks. I actually believed it and went from an FX8350 to i7 3770K. Did not notice a huge increase in frames ... in fact it was single digits

Anyway I recently built an i5 pc for a friend and compared it against the FX8350

Thief on i5 was 51.1 (highest settings) and on the FX it was 43.9. That was the biggest difference

Tomb Raider was 63.9 on the i5 and 60 on the FX

Rubbish gaming CPU ? I don't think so

Not totally rubbish but lower performing than Intel with a higher power consumption for the same money.

 

[MiddleOfTheRoad]

Someone may find them usable in GPU limited games but not every game is GPU limited, and the big problem with the FX8320 or 50 is that they are inconsistent.

It's not the hardware that's inconsistent, its the software. It's not like an FX-8320 or 50 chip gets asthma or something running a game. The games that exhibit that performance boost are optimized for hyperthreading / compiled for Intel processors. For actual performance of the hardware, Passmark is pretty much spot on that an FX-8350 is a hair slower than an i7 3770.

Now that a heck of a lot more games are being developed for AMD CPU's (thanks to Xbox One and PS4 design wins) -- I doubt that lopsided performance for games will be as noticeable in the future.

 

[MiddleOfTheRoad]

Not totally rubbish but lower performing than Intel with a higher power consumption for the same money.

Depends on the task. Playing games? Sure.

Multi-threaded apps -- an FX 8350 is probably faster than an i5 at just about all of them.

 

[MiddleOfTheRoad]

Wow. It is hard to believe that a shared FPU in itself can possibly cause delays of that magnitude. To go from 500 on an i5 to over 12000 on the AMD tells me that there has to be some serious cache/memory thrashing or something on that order of magnitude. This is exactly the kind of obvious bug that could be profiled out, if AMD had any money to spend.

It's hard to believe that programmers would release a game so poorly optimized. Funny how the shared FPU doesn't cause these delays on 99.999% of the other games on the market.

 

[Phynaz]

It's not the hardware that's inconsistent, its the software. It's not like an FX-8320 or 50 chip gets asthma or something running a game. The games that exhibit that performance boost are optimized for hyperthreading / compiled for Intel processors. For actual performance of the hardware, Passmark is pretty much spot on that an FX-8350 is a hair slower than an i7 3770.

Now that a heck of a lot more games are being developed for AMD CPU's (thanks to Xbox One and PS4 design wins) -- I doubt that lopsided performance for games will be as noticeable in the future.

So it's the software's fault that it doesn't run well on AMD, not AMD's fault that the CPU doesn't run software well.

BTW, the CPU in the game consoles isn't an FX chip, so optimizing for a console doesn't get the games optimized for a PC.

 

[Phynaz]

Depends on the task. Playing games? Sure.

Multi-threaded apps -- an FX 8350 is probably faster than an i5 at just about all of them.

Probably? Guess what, the very site that you're posting on does CPU reviews. An 8350 loses to an i5 in almost every application. http://www.anandtech.com/bench/product/697?vs=1261

We won't even talk about an i7.

 

[ViRGE]

This isn't an AMD vs. Intel thread. Fix the OP's problem or stay out.
-ViRGE

 

[Blue_Max]

If the spiking means everything running fine, freezing for a split second (or more), then resuming normal activity; I don't think it's a CPU issue... more like the motherboard chipset pausing to load data from the hard drive. (Guessing, but I've sure seen it on my machines!)

 

[Gikaseixas]

I don't think it's a CPU issue... more like the motherboard chipset pausing to load data from the hard drive

I've said before that AMD's main problem is not the CPU as I believe the FX 8xxx and 9xxx series are very good CPUs actually. The issue is the ancient motherboard chipsets with 2nd rate SATA and USB performance.
Every activity that depends mainly on the CPU, my FX 8350 excels but If I compare memory, SSD and USB speeds vs my Intel system, the performance in night and day.

 

[monstercameron]

did the OP ever find out what was wrong?

 

[guskline]

I've said before that AMD's main problem is not the CPU as I believe the FX 8xxx and 9xxx series are very good CPUs actually. The issue is the ancient motherboard chipsets with 2nd rate SATA and USB performance.
Every activity that depends mainly on the CPU, my FX 8350 excels but If I compare memory, SSD and USB speeds vs my Intel system, the performance in night and day.

Excellent point. My 3930k and 3770k machines are faster than my FX8350 machine. However, using ssd's for the OS on each and a solid 7200 rpm secondary drive sure helps. My FX8350 is using the Asus Sabertooth 990FX chipset rev 1. Since the cpu is custom water cooled, I can crank the cpu to 4.7Ghz (manual voltage 1.481 in BIOS) and temps are good and stability good when running all the benches including prime etc. Moreover, using a ssd for the OS coupled with a fast secondary drive helps to compensate for an older chipset.

I think the culprit maybe the mb/chipset and hdd or lack of ssd.

The 8320/8350 and uber 9 series really need the power/voltage to perform (the 8150 was worse- I had one.)

 

[MiddleOfTheRoad]

So it's the software's fault that it doesn't run well on AMD, not AMD's fault that the CPU doesn't run software well.

BTW, the CPU in the game consoles isn't an FX chip, so optimizing for a console doesn't get the games optimized for a PC.

And yes. Everytime I run a pure synthetic benchmark under Linux -- the i7 3770k and FX-8320 are just about side-by-side in virtually every performance metric. They score within 100 points of each in the BOINC benchmark for floating point (2600 vs 2500 pts).

The Intel Compiler cripples AMD chips.... and AMD has even sued
Intel over it. It's well documented.
http://www.agner.org/optimize/blog/read.php?i=49#49

Vishera and Jaguar do share commonalities for instruction sets -- If you optimize for an AMD APU, you should see a solid bump on the FX.

Its the software, not the hardware.

Disregard mod orders at your own peril. When we say this isn't an AMD vs. Intel thread we mean it
-ViRGE

 

[MiddleOfTheRoad]

I've said before that AMD's main problem is not the CPU as I believe the FX 8xxx and 9xxx series are very good CPUs actually. The issue is the ancient motherboard chipsets with 2nd rate SATA and USB performance.
Every activity that depends mainly on the CPU, my FX 8350 excels but If I compare memory, SSD and USB speeds vs my Intel system, the performance in night and day.

Totally agree with that -- AM3 is an ancient platform and it's shocking that you can still buy new chips for it.

 

[BSim500]

[snip]

Back on topic : The problem with OP's frame-rate latency with the FX-8350 is well documented and has been benchmarked on other tech sites and unfortunately isn't a one-off aberration or "broken motherboard". This is especially true of some of his listed games. Eg:-

This is very obviously a core saturation issue. Whilst Skyrim does scale to 4 cores to some degree, it doesn't scale beyond that (nor do most other games except for a recent few). It also needs high IPC on those cores simply because Skyrim is one of those games that just needs it. If Cinebench is anything to go by, Intel's are least 40-50% faster per Ivy core (and nearer 55-62% faster per Haswell core) even with a 500-600MHz clock disadvantage, ie, each 4Ghz FX-8350 core is individually about the same speed as a 2.5-2.6GHz Intel Haswell i5/i7 core, and as a result ends up stuttering far earlier / more often on certain games.

What's the solution for OP? Ocing a bit may help but won't close a 9.5x gap. Really it's just plain down to the fact not all games are heavily threaded. Even in "next gen" games often it varies across the map (ie, "quiet" areas may be threaded less well than "busy" areas even within the same game / map). Same with Skyrim with benchmarking inside a large city vs inside an empty cave. On games that are very CPU heavy but use only 2-4 cores, AMD's cores are simply being saturated earlier & more often. Beyond that there's really not much you can do as Skyrim just needs fast cores (as did predecessors Oblivion & Morrowind where the difference is even more pronounced due to using only 1-2 cores).

 

[davie jambo]

The skyrim thing you posted up there is out of date

The game used X87 instructions , AMD chips don't like that

They (Bethesda) fixed it

I played through it , ran 60fps on my FX8350 and 7970

Did not drop one single frame

 

[JimKiler]

I think the OP has a bad part, GPU, CPU or mobo. I had a Gigabyte mobo for my previous AMD Phenom II X4 CPU and everytime i isntalled Win7 it took 10 times longer to boot the CD and load the initial windows files. Once i upgraded only the mobo my issue was gone. I am not saying the mobo is the issue but it very well could be.

But i see no reason at the low resolutions being used an SSD would cause the piss poor performance the OP is seeing.