AMD's dx 11 CPU overhead is essentially gone (at least on Polaris)

[Nena Trinity]

3 minutes ago, Droidbot said:

You realise that the FX processor line is 6 years old, right? The 8350 bottlenecks the RX 480 and 1060 by a good margin even compared to a 26/2700k...

1 minute ago, Majestic said:

Even in low overhead scenario's the FX series loses to the new G4560. So i'm not sure what longetivty you're expecting out of a budget-level chip at this stage.

 

It's mostly Polaris though, so probably not.

Yes but its MC score is equal to a i5-6600K...

 

[Majestic]

Whatever, this isn't my topic and I'm not going to derail it with another pointless FX vs. Intel discussion.

[Nena Trinity]

3 minutes ago, Majestic said:

Whatever, this isn't my topic and I'm not going to derail it with another pointless FX vs. Intel discussion.

That is not the point, the point is a older i7 or Xeon would also do the job!

[Dabombinable]

47 minutes ago, Nena360 said:

About time! (now I can buy Vega for my FX-8370!)

How about you replace that CPU with something better. As it is, there are few tasks where a Skylake or Kabylake i3 isn't better than it-and in the tasks an FX 8370 pulls ahead, it only matches a locked i5....

 

Edit: BTW, see how it performs in games when you disable 1 core/ALU in each module-it should run better in games as the cache isn't being fought over any more.

[Nena Trinity]

1 minute ago, Dabombinable said:

How about you replace that CPU with something better. As it is, there are few tasks where a Skylake or Kabylake i3 isn't better than it-and in the tasks an FX 8370 pulls ahead, it only matches a locked i5....

Nope wont do that until it dies! ((I will probably buy the lower end Vega if I do not wait for the next gen after that tough I am planning on buying a GPU near GTX 1070 range i terms of performance) But my 880K on the other hand which will get a GTX 970 will be replaced next upgrade after that in favor of a RyZen or Intel CPU tough I am not sure if it will be a 8-core or 6-core!

[TheRandomness]

7 minutes ago, Dabombinable said:

How about you replace that CPU with something better. As it is, there are few tasks where a Skylake or Kabylake i3 isn't better than it-and in the tasks an FX 8370 pulls ahead, it only matches a locked i5....

 

Edit: BTW, see how it performs in games when you disable 1 core/ALU in each module-it should run better in games as the cache isn't being fought over any more.

That'd make for an interesting test, an i5-4460/6400/7400 vs. Half an 8350 at the same clocks to see what would happen if the cores didn't have to fight over cache.

[Dabombinable]

2 minutes ago, Nena360 said:

Nope wont do that until it dies! ((I will probably buy the lower end Vega if I do not wait for the next gen after that tough I am planning on buying a GPU near GTX 1070 range i terms of performance) But my 880K on the other hand which will get a GTX 970 will be replaced next upgrade after that in favor of a RyZen or Intel CPU tough I am not sure if it will be a 8-core or 6-core!

An i5 6500 is better in all tasks than your 8370 while consuming less power and dumping less heat into your room (just like my Xeon X5450 and Core 2 uad X6850-you really don't want to run it in a small or closed room, they can make effective heaters.

And I was thinking along the same lines with my i5 4440....but when it started bottlenecking my GTX 970 (just like the FX 8*** series does), I replaced it because there is no way in hell I'm going back to any experience similar to the stuttery mess that was my Core Duo T2600+uadro NVS110M/Geforce 7200Go. Also, your kind of nuts if you intend to run a GTX 1070 on your FX 8370....

[LeStringMan]

3 minutes ago, TheRandomness said:

That'd make for an interesting test, an i5-4460/6400/7400 vs. Half an 8350 at the same clocks to see what would happen if the cores didn't have to fight over cache.

In gaming performance? Makes literally no difference, the cpu only has 4 floating points anyways, and most games use the fpus instead of alu.

[Nena Trinity]

1 minute ago, Dabombinable said:

An i5 6500 is better in all tasks than your 8370 while consuming less power and dumping less heat into your room (just like my Xeon X5450 and Core 2 uad X6850-you really don't want to run it in a small or closed room, they can make effective heaters.

And I was thinking along the same lines with my i5 4440....but when it started bottlenecking my GTX 970 (just like the FX 8*** series does), I replaced it because there is no way in hell I'm going back to any experience similar to the stuttery mess that was my Core Duo T2600+uadro NVS110M/Geforce 7200Go. Also, your kind of nuts if you intend to run a GTX 1070 on your FX 8370....

I am sure as long as it can max out all games with my custom settings and give me 60+ FPS in all titles it is good to go! Also a i5-4440 gave you a stutters? I mean if it was a GTX 1070 on GTA V and you aimed for 144fps then sure but cmon it most been able to do 60fps just fine when even my 880K manages to max out a GTX 960 in all my games! o3o ((also no I am not nuts, I suspect you just suck at PCs))

[Dabombinable]

On 2/1/2017 at 6:43 PM, TheRandomness said:

That'd make for an interesting test, an i5-4460/6400/7400 vs. Half an 8350 at the same clocks to see what would happen if the cores didn't have to fight over cache.

Looking around I'm seeing nothing concrete as to how much the performance is increased-so until people actually start doing tests and sharing results, I think it will remain a hypothesis.

On 2/1/2017 at 6:48 PM, Nena360 said:

I am sure as long as it can max out all games with my custom settings and give me 60+ FPS in all titles it is good to go! Also a i5-4440 gave you a stutters? I mean if it was a GTX 1070 on GTA V and you aimed for 144fps then sure but cmon it most been able to do 60fps just fine when even my 880K manages to max out a GTX 960 in all my games! o3o ((also no I am not nuts, I suspect you just suck at PCs))

Yep. And it first started having issues in GTA V, then Fallout 4-and its single threaded speed is the same as an FX 8350 at 5.4GHz. Also, a GTX 960 is a lot weaker than a GTX 970. And how do I suck at PCs when you deny what has been proven time and again.

[Droidbot]

1 minute ago, Dabombinable said:

Yep. And it first started having issues in GTA V, then Fallout 4-and its single threaded speed is the same as an FX 8350 at 5.4GHz. Also, a GTX 960 is a lot weaker than a GTX 970. And how do I suck at PCs when you deny what has been proven time and again.

 

I'm having a slight bottleneck at 2560x1080p in some games with a i5 4670k and a 1070. I'd hate to see it with a very antiquated 8350.. 

[Nena Trinity]

3 minutes ago, Dabombinable said:

Yep. And it first started having issues in GTA V, then Fallout 4-and its single threaded speed is the same as an FX 8350 at 5.4GHz. Also, a GTX 960 is a lot weaker than a GTX 970. And how do I suck at PCs when you deny what has been proven time and again.

Let me guess you like a idiot enabled the settings in GTA V that causes much more of a load on 1cpu core yet makes the game look not much better or any better at all? And Fallout 4 I have never had issues with that how the hell did you have issues with Fallout 4? I am sorry but you most have had a faulty i5 or something... o3o

2 minutes ago, Droidbot said:

I'm having a slight bottleneck at 2560x1080p in some games with a i5 4670k and a 1070. I'd hate to see it with a very antiquated 8350.. 

That CPU has even less total CPU horse power than both FX-8370 & i5-6600K silly, but just wondering what game was that?  

[LeStringMan]

3 minutes ago, Dabombinable said:

Looking around I'm seeing nothing concrete as to how much the performance is increased-so until people actually start doing tests and sharing results, I think it will remain a hypothesis.

I assure you it makes no difference, i also assure you there are many threads on this exact topic, because after the launch of Bulldozer some people thought running 1 core per unit would magically improve ipc.

[Nena Trinity]

1 minute ago, LeStringMan said:

I assure you it makes no difference, i also assure you there are many threads on this exact topic, because after the launch of Bulldozer some people thought running 1 core per unit would magically improve ipc.

How did people do that? On my board you can only disable the whole module... (which is 4)

[Droidbot]

4 minutes ago, Nena360 said:

Let me guess you like a idiot enabled the settings in GTA V that causes much more of a load on 1cpu core yet makes the game look not much better or any better at all? And Fallout 4 I have never had issues with that how the hell did you have issues with Fallout 4? I am sorry but you most have had a faulty i5 or something... o3o

That CPU has even less total CPU horse power than both FX-8370 & i5-6600K silly, but just wondering what game was that?  

 

Yet, when paired with a 970, it gets a higher average and minimum framerate.. 

 

[TheRandomness]

9 minutes ago, Dabombinable said:

Looking around I'm seeing nothing concrete as to how much the performance is increased-so until people actually start doing tests and sharing results, I think it will remain a hypothesis.

I'm very tempted to get a 6300 and a cheap motherboard to support it, then run the 6300 at 4GHz with half of each module disabled, then compare to my i7 with hyperthreading off and one core disabled

[LeStringMan]

1 minute ago, Nena360 said:

How did people do that? On my board you can only disable the whole module... (which is 4)

ASRock and Gigabyte boards will let you run one core per module, not sure about Asus and MSI, it is in the core control section where you can also disable modules.

[Nena Trinity]

3 minutes ago, Droidbot said:

Yet, when paired with a 970, it gets a higher average and minimum framerate.. 

 

Those games are old now and none of them are properly optimized for more than 2 or 4 cores...

2 minutes ago, LeStringMan said:

ASRock and Gigabyte boards will let you run one core per module, not sure about Asus and MSI, it is in the core control section where you can also disable modules.

Well my FM2+ board lets me do that but not my AM3+ one and both are from MSI!

2 minutes ago, TheRandomness said:

I'm very tempted to get a 6300 and a cheap motherboard to support it, then run the 6300 at 4GHz with half of each module disabled, then compare to my i7 with hyperthreading off and one core disabled

No that will be no good unless it is a 2000/3000 series i7!

[TheRandomness]

3 minutes ago, Nena360 said:

Those games are old now and none of them are properly optimized for more than 2 or 4 cores...

Well my FM2+ board lets me do that but not my AM3+ one and both are from MSI!

No that will be no good unless it is a 2000/3000 series i7!

Well, since allrecent Intel architectures are within 10% of each other, it'd give a decent (but rough) comparison. Also, i7 would be clocked at 4GHz too c:

[Nena Trinity]

1 minute ago, TheRandomness said:

Well, since allrecent Intel architectures are within 10% of each other, it'd give a decent (but rough) comparison. Also, i7 would be clocked at 4GHz too c:

I want to see the result tough try the test on the FX chip with turbo disabled(locked to 4GHz) and enabled...

[Dabombinable]

46 minutes ago, Nena360 said:

Let me guess you like a idiot enabled the settings in GTA V that causes much more of a load on 1cpu core yet makes the game look not much better or any better at all? And Fallout 4 I have never had issues with that how the hell did you have issues with Fallout 4? I am sorry but you most have had a faulty i5 or something... o3o

That CPU has even less total CPU horse power than both FX-8370 & i5-6600K silly, but just wondering what game was that?  

I just maxed everything out - and why would I set GTA V to load a single core? I'm frustrated as it that Skyrim switches between using 4 cores evenly and 1 core heavily (the FPS suddenly drop+it stutters badly). And faulty i5? PMSL, actually doing a comparison with my 4790K downclocked with HT disabled, it turns out that its performance was actually normal, also-face it, single threaded an FX 8350 is terrible (and its multi threaded scaling as well).

[Dabombinable]

55 minutes ago, TheRandomness said:

I'm very tempted to get a 6300 and a cheap motherboard to support it, then run the 6300 at 4GHz with half of each module disabled, then compare to my i7 with hyperthreading off and one core disabled

I almost got an FX 8350+Sabertooth 990FX for $100 a year or 2 ago. But I already had too many spare motherboards+CPU (still trying to cut back).

[-rascal-]

@ivan134 This is interesting...

I'll be sure to do some testing on my end.

 

I'm personally on driver 16.12.2 since I built my X99 system, but driver uninstall/install will be quick.

I still have my FX-8350 system here with my 2nd R9-Fury still in there -- do some comparisons on both platforms.

 

Weekend activity scheduled.

 

[Clanscorpia]

Ill update and check OW, dont have GTA, but even after 1 update I saw like 30 FPS more in OW but Ill check w/ this update

[Humbug]

7 hours ago, Nena360 said:

Snip

it's worth noticing in those gaming benchmarks how much better piledriver fx-8350 is than bulldozer fx-8150. Quite a leap for a closely related CPU that came only 1 year later.