[MindBlankTech]The i3 7350K vs the FX-8300 (8350/8370) - how do they stack up?

[PontiacGTX]

 

[superstition]

update:

Here is pricing comparison for those with Micro Center available:

The AsRock board should manage 4.5 but it will need a good cooling fan on the VRM sink and airflow to the socket.

Alternatively, one could spend more and get an Aura

For those with a Micro Center, an i3 7350K is dramatically more costly.

------------------------------- original post -------------------------------

I see he overclocked the FX to 4.5 I hope he disabled APM and wasn't throttling. A Biostar 970 board is an interesting choice. Does it even have a heat sink on the VRMs?

As for Battlefield being the biggest difference it's not surprising considering who the developer is.

It would have been nice to see 1440 or higher resolution testing. Some testing has shown the FX to stand up better versus Intel at higher resolutions because of the GPU being a bigger factor.

 

[Crumpet]

I was watching on my phone so the audio was crap, why did he not test dx12?

Also his choice of amd components is... odd

 

[superstition]

I was watching on my phone so the audio was crap, why did he not test dx12

Also his choice of amd components is... odd

He did DOOM with Vulkan. Deux Ex MD, GoW4, and RoTR were tested with DX12.

Aside from the motherboard it looks solid enough. 1866 CAS 9 RAM is fine and he didn't use the stock cooler for his overclock.

If he is using that Biostar board and not the AsRock, though, I'd be definitely concerned about VRM throttling at 4.5. I'd also be concerned with the Biostar board. 4.5 GHz doesn't sound like much but when you're dealing with an 8 core CPU at 32nm the VRM system needs to be decent and, especially, needs to have active cooling. I don't know if he put a fan onto the VRM sink or not. Throttling VRMs certainly won't help performance.

 

[PontiacGTX]

I see he overclocked the FX to 4.5 I hope he disabled APM and wasn't throttling. A Biostar 970 board is an interesting choice. Does it even have a heat sink on the VRMs?

As for Battlefield being the biggest difference it's not surprising considering who the developer is.

It would have been nice to see 1440 or higher resolution testing.

it would be GPU limited then

If he is using that Biostar board and not the AsRock, though, I'd be definitely concerned about VRM throttling at 4.5. I'd also be concerned with the Biostar board. 4.5 GHz doesn't sound like much but when you're dealing with an 8 core CPU at 32nm the VRM system needs to be decent and, especially, needs to have active cooling. I don't know if he put a fan onto the VRM sink or not. Throttling VRMs certainly won't help performance.

IF would have the minimum FPS really low and AVG lower than they are, but there is a chance that it could be not really 100% CPU Bound for example in Crysis 3

 

[superstition]

it would be GPU limited then

Every system balances price and performance. If you own a 1440 or 4K monitor and want to know if it's a good value to get one or the other who cares if it's GPU limited. These are not synthetic benchmarks. They're practical ones.

Knowing to what extent the games would be GPU limited is very useful.

There might also be architectural differences that might help one architecture do better at a high resolution. As I recall, the 8 core Haswell fared better than the FX at lower resolutions but the FX could beat it at 4K. This is a dim recollection, though.

 

[PontiacGTX]

Every system balances price and performance. If you own a 1440 or 4K monitor and want to know if it's a good value to get one or the other who cares if it's GPU limited. These are not synthetic benchmarks. They're practical ones.

Knowing to what extent the games would be GPU limited is very useful.

the difference would be smaller unless games is multithreaded, then the G4560 would perform as bad regardless resolution

 

[superstition]

the difference would be smaller unless games is multithreaded

You're saying the GPU would make the delta between the CPUs smaller at higher resolutions and settings? Ok. And?

One thing I remember is how the Anniversary Pentium did OK at lower quality settings and resolutions (using a 750 Ti) and then, paired with a better GPU, stuttered all over in the Eurogamer tests in Crysis 3 and BF4. Architectural difference...

 

[superstition]

It's good that the reviewer did a 4.5 overclock on both. That's sensible. No one buying either of those chips today is likely to sink a lot into equipment to get a higher overclock than that.

Micro Center is the place to buy FX bundles, though, not Newegg.

 

[PontiacGTX]

You're saying the GPU would make the delta between the CPUs smaller at higher resolutions and settings? Ok. And?

Makes the CPU comparison pointless and hides the bottleneck

One thing I remember is how the Anniversary Pentium did OK at lower quality settings and resolutions (using a 750 Ti) and then, paired with a better GPU, stuttered all over in the Eurogamer tests in Crysis 3 and BF4. Architectural difference...

if the CPU would be in multiplayer in BF4 or some CPU bound scene in Crysis 3 it would have gotten stutering anyway

 

[Crumpet]

So let's get this straight;

Intel i3 build;
Team Elite Plus 16gb DDR4 2400 - $90
Gigabyte GA-Z270P-D3 - $109
i3-7350K - $180

Total - $389

and he's comparing it to;

AMD fx8300 build;
G Skill 16gb DDR3 1866 - timings 10-11-10-30 - $98
Biostar TA970 - $50
Fx-8300 @$115

Total - $263

So we've got a bad motherboard, overpriced ram and a bad choice of AMD processor.

Using the same website if he'd compared it to;

Team Dark 16GB 1600 - timings 9-9-9-24 - $97
ASrock Fatal1ty 990FX - $120 ($90 after rebate)
Fx-8320 @$125 ($10 off with promo code, $115)

Total - $302 after rebate and promo and a much fairer test. 1800 ram with CAS 10 in theory is slower in most games than 1600 CAS 9, the 990fx board will be better for overclocking and the fx-8320 is a better processor

 

[superstition]

1800 ram with CAS 10 in theory is slower in most games than 1600 CAS 9, the 990fx board will be better for overclocking and the fx-8320 is a better processor

He used 1866 CAS 9 as I already posted. If you're running 1600 you want to run 8-8-8, 8-9-8 at the worst.

8320 is irrelevant. All those CPUs perform the same when statically overclocked. The only thing to watch out for is the 9000 series which has poor leakage characteristics.

990 boards are only needed if one plans to run dual GPUs. I can find a board that will do 4.5 GHz for less than $90. Micro Center is really the place to go if one has one in the area. The Aura board plus the 8320E was as low as $140 plus tax, combined, after rebate.

 

[superstition]

Makes the CPU comparison pointless and hides the bottleneck

Hardly pointless when the bottleneck doesn't exist.

 

[Crumpet]

http://www.gskill.com/en/product/f3-1866c10d-16gab

Check the latencies. 10-11-10-30-2N

990FX boards tend to have more power phases, better vrm cooling etc.

 

[superstition]

http://www.gskill.com/en/product/f3-1866c10d-16gab

Check the latencies. 10-11-10-30-2N

990FX boards tend to have more power phases, better vrm cooling etc.

4.5, especially on an E processor, is not terribly demanding. 990 boards are not necessary to hit that level of overclock. The 970 Aura board has better VRMs than some 990 boards. It has a native 7+1 digital configuration.

The only reasons to get a 990 are:

1) run dual GPUs

2) buy a highest-end boards for high overclocking (Crosshair and Sabertooth mainly)

Doing #2 right now makes zero sense.

 

[superstition]

Updated my first post to include current pricing comparison, for those with a Micro Center available.


AMD CPU + board: $133

fancier AMD CPU + board: $161

Intel CPU + board: $273


AMD build is up to $140 cheaper, assuming one gets the same cooler for either build.

So, a person could buy two of those AMD combos for less than the price of one Intel i3 plus that board (the cheapest Z270 available at that Micro Center).

That also doesn't factor in the game code(s).

 

[Crumpet]

That biostar board has 4+1 and whilst it has heatsinks on the VRMs people don't recommend going over 1.35v on it. Personally, I think this poor is a poor choice for this comparison. If they wanted to use a 970 they could have used a GA-970-UD3P

-edit- either way, I think his video shows regardless that it's not really worth splashing out all that extra money on the i3 if you're going for a budget build.

 

[superstition]

That biostar board has 4+1 and whilst it has heatsinks on the VRMs people don't recommend going over 1.35v on it. Personally, I think this poor is a poor choice for this comparison.

Yeah, I said that in my opening post.

The UD3P will do 4.5 but it's more optimal for 4.4 because of its BIOS boot bug. Any multiplier beyond 4.4 GHz will cause it to fail to post so you have to use the BCLK to overclock which is less than optimal. It does work, though.

The UD3P is overpriced right now for what it offers. It's an outdated board. But, it's certainly true that it would have worked for a 4.5 GHz demo provided that the overclocker knows how to do BCLK overclocking and work around the other quirks of the board (needs plenty of CPU NB and strong airflow to the VRM sink).

 

[AtenRa]

Today, 3-4 weeks before ZEN launch you will be "naive" to buy any of those two.

 

[superstition]

More naive to pay twice the price for that i3 unless you're mainly doing Photoshop or something else that doesn't scale much.

I'll be surprised if Zen puts anything into $133 (CPU + board) territory. Micro Center might drop the price of the 8320E by $15 again. It was as low as $75 on BF.

 

[frozentundra123456]

That 7350k is the worst possible value choice. A lower level i3 or even the new hyperthreaded Pentium would make the price very close or possibly even lower for the Pentium. One must remember that a lot of users don't have access to a microcenter. Plus, if you want to base the decision on cost, an overclocked FX will use a lot more power than a Pentium or i3.

 

[ElFenix]

]Plus, if you want to base the decision on cost, an overclocked FX will use a lot more power than a Pentium or i3.

do you know how much electricity $140 buys? from anandtech's testing, overclocked to 4.5 ghz, the 8320e uses 120 watts more than the 7350k. lets assume 4 hours a day of doing that. so, every other day the 8320e will use 1 kwh more than the 7350k. at current US electric rates of 12.75 cents per kwh, that's nearly 2200 days difference. do you remember what hardware you were rocking 7 years ago? even accounting for air conditioning, you're talking about 5 years of difference.

 

[superstition]

One must remember that a lot of users don't have access to a microcenter.

I remember that which is why I say "If one has access to a Micro Center". Clearly, FX is generally less optimal when purchased elsewhere.

If one does, though, have access to Micro Center, then it becomes more compelling for some budgets.

I wonder what Frys has, also, in terms of bundle deals. I know, vaguely, that sometimes that company runs specials.

As for power usage, some people don't spend a huge amount of time gaming. Having a BIOS profile for less demanding work will enable an E chip to sip power.

 

[lopri]

Those two CPUs are at polar opposite ends as far as desktop CPUs are concerned. It is like this:

vs.

The 'tubers' motive, methodology, as well as the results are all questionable.

 

[superstition]

Those two CPUs are at polar opposite ends as far as desktop CPUs are concerned. It is like this:

The 'tubers' motive, methodology, as well as the results are all questionable.

Considering that the i3 costs twice as much if one has a Micro Center available I suppose. However, it's good for people to know what kind of performance they can expect if they go in one direction or the other. The thing that's most disappointing about his roundup is that he stuck with just 1080 resolution. A dual 480 test would have also been good because a lot of people scored those on sale on Black Friday for low cost.

At least he put both at 4.5. So many people dismiss Intel vs. AMD results because the Intel CPU isn't overclocked. I would have liked to have known that he used a capable board with VRM cooling for the AMD, though — and had APM turned off.