147 replies to this topic

[Bill Lumbar]

ShinobiHunter, on 24 March 2015 - 04:33 PM, said:

How long should I leave prime95 run before trying to up my multiplier again? I know some people recommend 8 hours, but that seems a bit excessive... What are your thoughts?

Right now I'm sitting at 4.5Ghz with core voltage at 1.432V. My CPU temp is staying right around 57-58 °C

True die harders say 24 hour min..... lol. I say go a couple hours and call it good. If it is unstable, you will find out soon enough in gaming.

[ShinobiHunter]

Cool, thanks. I really appreciate the help.

[Hardin4188]

When looking at amd fx temps you have to understand that they aren't 100% accurate. Essentially they are close estimates. You also have to keep in mind there are two temperatures you need to look at. The package temperature which is calculated from diodes on each corner of the chip and the socket temp, which is calculated from the motherboard diode.

[ShinobiHunter]

Socket temp is TMPIN temperatures, right? Or do I find that somewhere else?

[Hardin4188]

ShinobiHunter, on 24 March 2015 - 04:49 PM, said:

Socket temp is TMPIN temperatures, right? Or do I find that somewhere else?

On HWMonitor it is usually CPUTIN. You'll want to keep this below 72c and the package temp below 62c. These are the maxes that you shouldn't exceed in prime 95. If you are reaching these temps in the game then panic.

[Goose]

ShinobiHunter, on 24 March 2015 - 04:33 PM, said:

How long should I leave prime95 run before trying to up my multiplier again? I know some people recommend 8 hours, but that seems a bit excessive... What are your thoughts?

Are you setting it to Custom, and set to all-but 4GB of your memory?

When it crashes at a self-test of 24k or larger, as found in the "results.txt," then you crashed in the "LargeFFT" portion of the game, which means … things and stuff to the types of Intel chip that has a "Overclockes BSoD List" written for it. Less then 24k is a SmalFFT, and tends to mean just bump the vcore, as does any rounding error, or program lock.

Here ya' go: https://www.google.c...5.0.ndp8vt4mOZs

[ShinobiHunter]

I tried to get it to 4.6GHz but it wasn't stable until I bumped the voltage to 1.464V. That was getting a bit too toasty for my liking, the package temp hit 61 degrees and I dialed it back to 4.5GHz at 1.432V. So far Prime95 is stable.

[Lord Letto]

Hardin4188, on 24 March 2015 - 04:55 PM, said:

On HWMonitor it is usually CPUTIN. You'll want to keep this below 72c and the package temp below 62c. These are the maxes that you shouldn't exceed in prime 95. If you are reaching these temps in the game then panic.

?
with HWMonitor, i'm not seeing a CPUTIN, for Mobo (Gigabyte 990FXA-UD3 Rev. 1.1) it's TMPIN0, TMPIN1 & TMPIN2, it also gives me a "Package" Temp for my FX-4100, Air Flow Temp for 1 of my HDDs and a TMPIN0 for my HD 6970.

[Hardin4188]

Lord Letto, on 24 March 2015 - 06:28 PM, said:

?
with HWMonitor, i'm not seeing a CPUTIN, for Mobo (Gigabyte 990FXA-UD3 Rev. 1.1) it's TMPIN0, TMPIN1 & TMPIN2, it also gives me a "Package" Temp for my FX-4100, Air Flow Temp for 1 of my HDDs and a TMPIN0 for my HD 6970.

Yeah it depends on the motherboard, they all label them a different way. I believe TMPIN0 is motherboard, and TMPIN1 is socket, but I am not 100% on this. Package is the temperature measured from the cpu itself, anything like tmp or cputin is from the motherboard diode. I have an asrock 990fx extreme9 and it's cputin.

Edited by Hardin4188, 24 March 2015 - 06:56 PM.

[Smokeyjedi]

Its pretty straight forward. CPU the socket and the individual cores....... from top to bottom is the only consistent factor.

Edited by Smokeyjedi, 24 March 2015 - 08:42 PM.

[PheonixStorm]

I find it odd that an obsolete machine like mine seems to have the same or perhaps fewer issues than the newer systems. I am right at the low end of "recommended" specs. But anyway... What kind of timing does your RAM have, are you running a 64-bit OS? Those are two other things to watch for as well as HD speeds and the 12V rail on the power supply. The nice hidden services running in your OS could come back to haunt you as well if you aren't careful. Pretty much anything that runs in the background that sucks up CPU cycles as well as ram should be shutdown. If you are hurting for memory you should also switch your background to a solid color rather than that hi-res image of the purple haired anime chick

[NikkoChan]

hi-res image of the purple haired anime chick.... wait a min... I have a hi-res image of a purple haired anime catgirl! OH NOEs my memoriez are nomed

[Flapdrol]

PheonixStorm, on 24 March 2015 - 04:02 PM, said:

AMD are decent CPUs. I'm running a Phenom II X4 920 w/ 4GB ram and an old GTX 260 and have had no issues until after the patch on the 17th. Now I lag a bit on two maps but I still do very well. Just can't go max graphics (though honestly I haven't tried, may work but who knows).

It is true that Intel cpus are in some ways superior they are by far cheaper and on price just as good as intel. The problem is DIY builds are almost always superior to prebuilt system you can find online. Buying some companies computer comes at a price. They are bloated with useless software that takes up system resources and some of it is hard to get rid of.

Problem is the FX's aren't really an upgrade from the phenom II when it comes to most games, and this one in particular. AMD went for lower per core performance and more cores, and games are hard to effectively multithread.

As for stresstesting I'd recommend 8 hours of prime small fft (if al you did is increase cpu multi and vcore).

[Smokeyjedi]

Flapdrol, on 25 March 2015 - 02:12 AM, said:

Problem is the FX's aren't really an upgrade from the phenom II when it comes to most games, and this one in particular. AMD went for lower per core performance and more cores, and games are hard to effectively multithread.

As for stresstesting I'd recommend 8 hours of prime small fft (if al you did is increase cpu multi and vcore).

agreed. unless you are oveclocking to moderate levels like >4.8 the fx series wont surpass the Pii X4 @ 4.0ghz.......but talk about improved memory usage and all things performance generally speaking......

[Hardin4188]

The FX chips do let you use higher clocked memory and the cryengine 3 does like high memory bandwidth.

[Goose]

Hardin4188, on 25 March 2015 - 04:56 AM, said:

The FX chips do let you use higher clocked memory and the cryengine 3 does like high memory bandwidth.

You got a source for that? 'Cause I think so, but would like better proof then I have …

[xWiredx]

Hardin4188, on 25 March 2015 - 04:56 AM, said:

The FX chips do let you use higher clocked memory and the cryengine 3 does like high memory bandwidth.

Cryengine *might* but this game certainly does not seem to care. No fps difference between triple-channel DDR4-2400 and quad-channel DDR4-2750 (which is a pretty big difference in bandwidth). No perceivable difference between DDR3-1866 and DDR4-2400, either (from when I switched from my 2600K to my 5820K and kept the same GPU).

[Hardin4188]

Goose, on 25 March 2015 - 10:15 AM, said:

You got a source for that? 'Cause I think so, but would like better proof then I have …

xWiredx, on 25 March 2015 - 11:35 AM, said:

Cryengine *might* but this game certainly does not seem to care. No fps difference between triple-channel DDR4-2400 and quad-channel DDR4-2750 (which is a pretty big difference in bandwidth). No perceivable difference between DDR3-1866 and DDR4-2400, either (from when I switched from my 2600K to my 5820K and kept the same GPU).

I haven't seen any benchmarks for this game with different memory speeds, so this is all based on what I've seen from Crysis 2 and 3 benchmarks. I will try to find some actual benchmarks showing this, but I'm about to leave work so I'll have to try and post it later.

I'm sorry I can't find the page where I originally saw it. From what I remember they were only a few frames different anyway. Nothing significant. If I do find anything I will definitely post it here.

Edited by Hardin4188, 25 March 2015 - 02:46 PM.

[Bill Lumbar]

Ok guys..... just got the chance to play tonight on the Phenom II 960T build.

Specs,

Phenom II 960T @ stock 3ghz, 3.3ghz turbo core (Thats right, no OC atm)

XFX 6870 1 gb card (for some reason I was thinking it was a 2gb, its not)

standard WD 7200 Black 16 meg cach drive

8 gb of kingston hyper X blue 1600hz

55" Lg Led tv, 1080P all settings are on medium, except object detail, very high, terrain detail set on High. DX11, damage glow on, post AA on, running 64 bit client.

Here is my first game benched and played.

2015-03-27 21:02:22 - MWOClient
Frames: 5514 - Time: 180000ms - Avg: 30.633 - Min: 18 - Max: 62 The bog

2015-03-27 21:06:24 - MWOClient
Frames: 6658 - Time: 180000ms - Avg: 36.989 - Min: 21 - Max: 55 The bog watching after death


2015-03-27 22:26:09 - MWOClient
Frames: 6772 - Time: 180000ms - Avg: 37.622 - Min: 25 - Max: 55 Caustic Valley


So far game play seems really off after coming from my 32" BenQ at 1440P and the I7 build. I would not call it a slide show yet by any means on this rig, but first impression is not what I am used to with my main gaming rig to say the very least.

CW drops.....


2015-03-27 21:33:08 - MWOClient
Frames: 7011 - Time: 180000ms - Avg: 38.950 - Min: 25 - Max: 71 CW Attack on Sulfurous

2015-03-27 21:39:47 - MWOClient
Frames: 5950 - Time: 180000ms - Avg: 33.056 - Min: 20 - Max: 54 CW Attack on Sulfurous

2015-03-27 21:43:18 - MWOClient
Frames: 5562 - Time: 180000ms - Avg: 30.900 - Min: 21 - Max: 49 CW Attack on Sulfurous

Edited by Bill Lumbar, 27 March 2015 - 06:35 PM.

[xWiredx]

Once you get a taste of the good life, it's weird spending a day with the norms.