Is it time to stop discouraging use of the FX-8350?

[runit3]

It's time to stop reopening this discussion because you're an expert who says an i7-5960x can't beat an i7-4770k just because?

I have an extremely hard time believing my 5960X at stock speeds would edge out any of the 4770/4790k's I've had my hands on, especially at 1080p. Maybe when the PG279Q drops I can actually do a decent comparison between my 4690k and the new 5960X I just got back.

 

Supposing that all of those gaming graphs actually are 100% accurate, it still begs the question why recommend a dead power hungry platform with few modern features? If the main basis of the argument is that cores are mattering more in games, the 5820k and whatever Skylake-E/Zen turn out to be should be the go-to recommendation -not a platform from 2012.

[Majestic]

Your alleged common sense is bullshit if it's being refuted by actual stats.

 

By a singular source of information, that is not being reflected through multiple websites. It's an anomaly that fits your theory and thus must be true. Utter, utter confirmation bias.

 

If you're going to judge or interpret benchmarks and wonder how a 5960X might possibly be in a disadvantage in gaming due to poor core-scaling, I fear you lack the know-how to do that job properly. Sorry man, but you're way out of line here.

[TheCaptain53]

I do think these results show an FX-8350 could be reasonable in budget builds if this very recent trend continues. It's funny how often people recommend against i7s and enthusiast cpus like the 5820k here as if all debate should start and end at the 4460/4690k for gaming.

That's the way it was/has been for a long time. People recommend the 4690K because for the price, it WAS the top dog. If/when more benchmarks show a heavy inclining to the 4790K and 5820K, more people are going to recommend them instead.

[SteveGrabowski0]

Go look at other websites that also use new games.

 

What other sites do good gaming benchmarks for cpus? By good I mean with maxed settings on overkill gpu setups to stress the cpu and make it as much of a bottleneck as possible. I can't read polish so I don't use pclab.pl's reports, and they don't seem to benchmark all new releases like GameGPU.ru does.

[Dabombinable]

What other sites do good gaming benchmarks for cpus? By good I mean with maxed settings on overkill gpu setups to stress the cpu and make it as much of a bottleneck as possible. I can't read polish so I don't use pclab.pl's reports, and they don't seem to benchmark all new releases like GameGPU.ru does.

Been around since the late 90's: http://www.anandtech.com/

http://www.techpowerup.com/

 

those are just a couple

[SteveGrabowski0]

By a singular source of information, that is not being reflected through multiple websites. It's an anomaly that fits your theory and thus must be true. Utter, utter confirmation bias.

 

If you're going to judge or interpret benchmarks and wonder how a 5960X might possibly be in a disadvantage in gaming due to poor core-scaling, I fear you lack the know-how to do that job properly. Sorry man, but you're way out of line here.

 

I use that source because they're the only one I know that does CPU tests for all major releases and actually cranks the settings with overkill gpu setups to force cpu bottlenecks. I don't know that pclab.pl does all major releases and I can't read polish to search for their benchmarks for specific games.

 

I think it's funny you just disregard extra cores and a huge amount of cache.

[Majestic]

I can't read polish so I don't use pclab.pl's reports

 

You can read russian then? Because Gamegpu is russian..

 

 

I think it's funny you just disregard extra cores and a huge amount of cache.

 

Which you provide no proof of that it has a huge impact. 4.2ghz vs 3.0ghz will have a bigger impact than a few MB of cache.

[juretrn]

Att this point, the 8350 is just a bad joke.

[SteveGrabowski0]

Been around since the late 90's: http://www.anandtech.com/

http://www.techpowerup.com/

 

those are just a couple

 

I use techpowerup all the time for gpu results, they're great there. Can you link to them reviewing cpu performance in the newest games though? I don't see it. When I looked up Witcher 3's performance article, it was all about gpus. And anandtech is reviewing every new game for cpu performance?

[Majestic]

 

 

See what I mean?

 

And then look at this;

 

One core maxed out. How the hell can the 5960X be higher.

 

I'm not saying with absolute confidence either is right. But the fact they contradict proves that you can't just blindly go by one and stick with it.

[SteveGrabowski0]

 

You can read russian then? Because Gamegpu is russian..

 

 

 

Which you provide no proof of that it has a huge impact. 4.2ghz vs 3.0ghz will have a bigger impact than a few MB of cache.

 

 

A 4770k runs at 3.7 GHz on full quadcore load at stock, the 5960x somewhere from 3.0 GHz to 3.5 GHz on all cores load (Intel didn't publish the turbo steps for the 5960x like they did for the 4770k).

[Ansau]

Answering the topic question: No.

- FX 8000 cpus belong to a quite old platform, so you end up buying old technology, which is ALWAYS a bad move.

- FX 8350 is already a quite overclocked cpu, and to squeeze more it requires a quite expensive MB and cooler, which goes contrary to the main point of AMD, prize. In the other hand, a Haswell i5 can be overcloked quite easily with a 100 bucks MB and a cheap cooler like a 212.

- Those comparisons are against the 4670k, when they should be with the 4690k, which runs 100MHz higher and can overclock a bit better.

- Comparing those games (with the 4690k in mind), the FX only wins in The Witcher, matches in 3 games and it loses in 2 more.

- Here you don't add other games released this year (GTA V, Arkham Knight, Project Cars, Ark: Survival Evolved, Total War: Arena...) in which the FX loses miserably. So it ends up winning only in 1, matching in 3 and losing in 7, at least. How can be this a good reason to encourage FX cpus?

- We have a couple future games using DX12 showing how punishing is for a cpu to have such a low single core performance, where a mere i3 can blow away a FX 8000.

[Nena Trinity]

Some games the FX is great for some its not and well we also have Arma which it does terrible in!

[runit3]

A 4770k runs at 3.7 GHz on full quadcore load at stock, the 5960x somewhere from 3.0 GHz to 3.5 GHz on all cores load (Intel didn't publish the turbo steps for the 5960x like they did for the 4770k).

http://www.hardwareluxx.com/index.php/reviews/hardware/cpu/32409-haswell-e-test-intel-core-i7-5960x.html?start=3

Only cores 0/1 run at 3.5, the other 6 at 3.3.

[Majestic]

Here, another one. Dying light, a game that ran on a single core almost because it was so poorly optimized for DX11 and pulled too many drawcalls.

 

 

 

So you'd think AMD hardware does poorly, and it did (i remember the rage when it launched). But by that same logic, how can a 3ghz ivy beat a higher clocked haswell. GameGPU's results make no sense, at all.

 

And if Cache was the reason, how come the i3 is higher than the 4670K which has 2MB more, and it squares with the i7-4770K which has double the cache.

 

Even the i3-2100 makes no sense, since it's IPC is higher than all AMD's in the line-up.

[SteveGrabowski0]

http://www.hardwareluxx.com/index.php/reviews/hardware/cpu/32409-haswell-e-test-intel-core-i7-5960x.html?start=3

Only cores 0/1 run at 3.5, the other 6 at 3.3.

 

Thanks for finding that. I would have seriously questioned the 5960x beating the 4790k so soundly in everything since I figured the 5690x would be around 3.2/3.3 at stock while the 4790k is 4.2 at stock. But since GameGPU uses a 4770k it definitely doesn't seem that crazy.

[Sauron]

BECOMING. It isn't right now, but when games fully take advantage I think we will really see big improvement on AMD cpus.

 

yeah, good luck with that. I bet it will be fun to have waited 6-7 years for good performance and being stuck with outdated technology in pretty much every respect, starting with pcie 2.0 and going down the list to ddr4 and m.2. And I also really don't believe it will get any better than this. i7s still beat fx8 cpus even in stuff like cinebench where all the cores are used to their full extent.

[SteveGrabowski0]

Here, another one. Dying light, a game that ran on a single core almost because it was so poorly optimized for DX11 and pulled too many drawcalls.

 

 

 

 

 

So you'd think AMD hardware does poorly, and it did (i remember the rage when it launched). But by that same logic, how can a 3ghz ivy beat a higher clocked haswell. 

GameGPU's results make no sense, at all.

 

I don't see any Ivy processor on that list, I'm not sure which result you're talking about. As for 5960x vs 4770k, it's a 400 MHz difference in clock. Maybe it's the cache making up the difference, there is a huge difference in cache sizes between the 5960x and 4670k/4770k. 

[Majestic]

I don't see any Ivy processor on that list, I'm not sure which result you're talking about. As for 5960x vs 4770k, it's a 400 MHz difference in clock. Maybe it's the cache making up the difference, there is a huge difference in cache sizes between the 5960x and 4670k/4770k. 

 

Oops my bad, the 5960X is haswell you're right, but it's still clocked significantly lower than the others and the game is clearly running on a single core almost. And the cache story doesn't fit either, check the other Haswell's for proof.

 

i3-4330 has 4MB, 4670K has 6MB and the 4770K has 8MB and they barely differ.

[SteveGrabowski0]

Oops my bad, the 5960X is haswell you're right, but it's still clocked significantly lower than the others and the game is clearly running on a single core almost. And the cache story doesn't fit either, check the other Haswell's for proof.

 

i3-4330 has 4MB, 4670K has 6MB and the 4770K has 8MB and they barely differ.

 

The 4330 and 4670k hit the same minimum though with the i3 clocked about 3% higher. That 2MB extra on cache on the 4670k could have contributed to that result considering yeah, that is one game that was completely dependent on single core performance at launch so I doubt it was the four physical cores vs two physical, two hyperthreads. That must have been a nightmare on AMD cpus until the 1.4 patch when loads started being spread across 4 threads (though the other 4 on my Xeon were still left close to idle most of the time when I played with the RTSS overlay on). Probably still terrible on AMD cpus since it doesn't scale to 8 threads at all.

[Majestic]

All that patch did was lower the draw distance by default. 

 

The i3-4330 is 3.5ghz, the 4670K is 3.6ghz boosted, probably higher because it's load is not even. So both the clockspeed and cache are higher.

 

It is within tolerance though, so it could just be variance in the testresults. But had cache made a detrimental impact, the i3 would've had to have been much lower on the scale.

I've also never seen a game really scale with cache, except for WoW.

[Prysin]

Eventhough i've used GameGPU's results in the past, they make no sense and as a result stopped using them. 5960X can never be above a 4770K in consoleports, not on 3ghz.

depends on the background tasks being used really..

if they use CPU intensive or poorly written programs that uses CPU cycles, then yes it can.

However i will say from these benchmarks, that it is now quite obvious that if you cannot manage to get AT LEAST a 4460, you should get a FX.

If nothing else, the FX is cheaper, will perform better then your alternative (i3) almost every time, not to mention, FX boards are really cheap, even OK ones.

While PCIe Gen3 may not be present, the ability to buy boards capable of pushing 2x 16x Gen2 is just as good as 2x 8x Gen3 (which is what most intel boards can offer for CF or SLI)...

but yeah. If you cannot afford an I5, intel isnt worth it. Upgrade path and all, with ZEN coming out (if it is really good, intel would prolly lower their prices a little) and time moving on as it always does, i see little reason or argument to EVER buy Intel below a i5 now...

For gaming, below i5, the benchmarks are VERY clear.

FX, even if it is old and has shit IPC, is better. It simply, just, fucking, is.

[Majestic]

If nothing else, the FX is cheaper, will perform better then your alternative (i3) almost every time, not to mention, FX boards are really cheap, even OK ones.

 

It's not cheaper, and it's not always better than an i3. Boards that support an FX 8 core are more epensive than H81/H110 boards that support haswell/skylake.

[Prysin]

It's not cheaper, and it's not always better than an i3. Boards that support an FX 8 core are more epensive than H81/H110 boards that support haswell/skylake.

oh please.

65 bucks for a board that can run a 8350 at stock 24/7

get me a decent intel board that can match the AMD boards features.

sure a B110 board is cheap...

then again the cheapest i3 for that board is 125 bucks. the cheapest DDR4 is 45 bucks (2133 CL15).

in comparison..

8350 is 140 ish bucks (cannot check atm. on mobile), board is 65, 8GB of 2400MHz CL10 RAM is 47 bucks...

sure, it costs more, and performs more.

a i3, even a skylake i3, cannot match the 8350 in games that uses more then 4 cores. it just doesnt. for that you need a locked i5.

now, i wouldnt reccomend then 8350 in particular, because of various reasons tied to FX, but the 6300 i would. It is by far cheaper then any i3, and given faster RAM, and a mild OC (4GHz-4.2GHz), and it will pretty much rival even the mighty i3 6100...

in games using all 6 threads, it should get close (not beat) a i5 4430 or 4440..

as for 8xxx series... id grab a 8320 and get a GA970a UD3P board so i could OC it to 8350 levels. stock cooler can handle 4GHz just fine as that is the normal turbo speed of FX8320.

[Majestic]

oh please.

65 bucks for a board that can run a 8350 at stock 24/7

get me a decent intel board that can match the AMD boards features.

sure a B110 board is cheap...

then again the cheapest i3 for that board is 125 bucks. the cheapest DDR4 is 45 bucks (2133 CL15).

in comparison..

8350 is 140 ish bucks (cannot check atm. on mobile), board is 65, 8GB of 2400MHz CL10 RAM is 47 bucks...

sure, it costs more, and performs more.

a i3, even a skylake i3, cannot match the 8350 in games that uses more then 4 cores. it just doesnt. for that you need a locked i5.

now, i wouldnt reccomend then 8350 in particular, because of various reasons tied to FX, but the 6300 i would. It is by far cheaper then any i3, and given faster RAM, and a mild OC (4GHz-4.2GHz), and it will pretty much rival even the mighty i3 6100...

in games using all 6 threads, it should get close (not beat) a i5 4430 or 4440..

as for 8xxx series... id grab a 8320 and get a GA970a UD3P board so i could OC it to 8350 levels. stock cooler can handle 4GHz just fine as that is the normal turbo speed of FX8320.

 

"Oh please" indeed, for you provide no evidence to your claims. Or produce any examples. Rather, i'm seeing the same dismissive claims being perpetuated that have been continuously shot down yet not registered by AMD pundits due to their severe cognitive dissonance. (not immediatly branding you one, but you're falling victim to their almost religious ideology).

 

 

 

i3-6100 does outperform the FX series in games that use more than 4 cores. And it can match a 2500K in games.

 

So let's look at a 8350 setup

PCPartPicker part list / Price breakdown by merchant

 

CPU: AMD FX-8350 4.0GHz 8-Core Processor  ($166.88 @ OutletPC) 

CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler  ($24.89 @ OutletPC) 

Motherboard: Gigabyte GA-970A-DS3P ATX AM3+ Motherboard  ($64.89 @ OutletPC) 

Memory: G.Skill Ripjaws X Series 8GB (2 x 4GB) DDR3-1600 Memory  ($44.99 @ Newegg) 

Total: $301.65

Prices include shipping, taxes, and discounts when available

Generated by PCPartPicker 2015-11-11 07:39 EST-0500

 

Cooler is mandatory, if you're going to argue the stock fan of the 8350 is "fine" we're done because you're disingenuous.

 

Now an i5 setup.

PCPartPicker part list / Price breakdown by merchant

 

CPU: Intel Core i5-6400 2.7GHz Quad-Core Processor  ($180.99 @ SuperBiiz) 

CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler  ($24.89 @ OutletPC) 

Motherboard: Gigabyte GA-H110M-A Micro ATX LGA1151 Motherboard  ($51.99 @ SuperBiiz) 

Memory: Kingston HyperX Fury Black 8GB (2 x 4GB) DDR4-2133 Memory  ($53.99 @ Newegg) 

Total: $311.86

Prices include shipping, taxes, and discounts when available

Generated by PCPartPicker 2015-11-11 07:45 EST-0500

 

Just for fun, I added the cooler anyway and it's still only 10 dollars more expensive. Without the cooler, which isn't really needed, it's cheaper. You can even squeeze in a faster model.

 

PCPartPicker part list / Price breakdown by merchant

 

CPU: Intel Core i5-6500 3.2GHz Quad-Core Processor  ($194.99 @ Amazon) 

Motherboard: Gigabyte GA-H110M-A Micro ATX LGA1151 Motherboard  ($51.99 @ SuperBiiz) 

Memory: Kingston HyperX Fury Black 8GB (2 x 4GB) DDR4-2133 Memory  ($53.99 @ Newegg) 

Total: $300.97

Prices include shipping, taxes, and discounts when available

Generated by PCPartPicker 2015-11-11 07:46 EST-0500

 

Motherboard features? The only benefit the 970A has are two extra Sata ports. And who uses 6 sata ports anyway these days for gaming machines.

 

I mean at this point I might aswell follow Faceman's example and copy this every time another one of these ill-researched posts gets posted.