possible that it's ok >70C AMD FX-8xxx??

[soccerballtux]

I'm still voting for this option. Or maybe an i3-4160/4170 if you play games on your PC. But if you don't do much multitasking (VMs, video editing, etc.), and you want high single-threaded performance (fast snappy web browsing), then the Pentium G3258 is your ticket. But if you really want to clock it, it appears that you need a Z97 board too.

no, G3258 is not acceptable, and it has minimum framerate issues. I'm not taking a step back in cores/threads. At the very least need 8

 

[Shehriazad]

no, G3258 is not acceptable, and it has minimum framerate issues. I'm not taking a step back in cores/threads. At the very least need 8

If you disable 2 modules...you only got 4cores/threads.

Bruh.


Anyway...I'd still agree on the G3258 matter...it's a step back since 2cores/threads already comes with a bunch of problems and annoyances nowadays.

 

[soccerballtux]

If you disable 2 modules...,you only got 4cores/threads.

Bruh.


Anyway...I'd still agree on the G3258 matter...it's a step back since 2cores/threads already comes with a bunch of problems and annoyances nowadays.

annnnnnd....that's twice as many asa the G3258

 

[2is]

annnnnnd....that's twice as many asa the G3258

Moar cores isn't better in this case. If you need at least 8 cores, why are you bothering disabling 4 of them?

 

[soccerballtux]

Moar cores isn't better in this case. If you need at least 8 cores, why are you bothering disabling 4 of them?

because 4 > 2

 

[Yuriman]

An i3 has something like 60% better IPC than an FX. Assuming a 3.8ghz i3 gets 30% extra multithreaded performance from HT, you'd need to reach ~6ghz on your FX chip to match its single-threaded performance, and 5ghz to match it with all 4 cores (2 modules) fully loaded. The extra threads the i3 has are a nearly ideal solution too, as the distribution of work across cores/threads is almost never even. Even with all 4 modules enabled, a 3.8ghz i3 would (hypothetically) match a 2.5ghz FX with all 8 cores maxed out in multithreaded workloads.

EDIT: And there are some workloads where an i3 gets significantly more than 30% from HT.

 

[2is]

because 4 > 2

Everything else being equal that would be true. Everything else isn't. 2 Intel cores > 4 FX cores. You don't have to believe it, but it won't make it any less true.

 

[Yuriman]

Everything else being equal that would be true. Everything else isn't. 2 Intel cores > 4 FX cores. You don't have to believe it, but it won't make it any less true.

Assuming 60% better IPC, and a 20% module penalty on Piledriver, you'd expect 2 Intel cores without HT to be approximately equal to 4 (equally clocked) FX cores when running 4 threads.

 

[soccerballtux]

Assuming 60% better IPC, and a 20% module penalty on Piledriver, you'd expect 2 Intel cores without HT to be approximately equal to 4 (equally clocked) FX cores when running 4 threads.

that checks out. the core i3 2100 at 4.5 something ghz does about 110 fps in handbrake android mp4 encode, and my CPU does ~220fps on the same at 4.3ghz.

 

[Yuriman]

Hyperthreading got a lot better with Haswell though, and there are 2 generations of IPC improvements on top of that. I expect Skylake i3's will beat FX-6xxx chips in perfectly threaded tasks.

 

[Ancalagon44]

Disabling 1 core per module instead of 2 modules would give you better performance.

 

[2is]

Hyperthreading got a lot better with Haswell though, and there are 2 generations of IPC improvements on top of that. I expect Skylake i3's will beat FX-6xxx chips in perfectly threaded tasks.

Not to mention Quick Sync

 

[soccerballtux]

..

 

[rancherlee]

Disabling 1 core per module instead of 2 modules would give you better performance.

I've always wondered about that one, I'd personally love to see a 2 tread/2 modules vs. 1 tread/4 modules type benchmark.

 

[soccerballtux]

I've always wondered about that one, I'd personally love to see a 2 tread/2 modules vs. 1 tread/4 modules type benchmark.

it's about a 10-15% speedup, I can ref this, it's been tested before. It'll take me a day to remember who and where the test occurred.