Discussion in 'General Discussion' started by Thomas Jansen, Jul 22, 2019.
Single channel memory might drop the performance a bit, but I doubt it will be as much as he gets.
No problem. I think most of us have fallen foul of the afterburner issue in the past.
Right, he lost about 80% performance so it cannot be the single channel slow memory.
@Norbi930523 As Raceroom is CPU bound and your CPU is not too bad, the slow single channel RAM though, limits the CPU.
2 sticks of 3200MHz in dual channel will give you some extra CPU performance.
Whats the Unigine score now after uninstalling afterburner and rivatuner ?
@Nir Tal This is my new Unigine performance after uninstalling Afterburner and Rivaturner (it's almost 4x improvement):
Not me, i use it to undervolt my GPU and set a fan curve, get my OSD and cap my FPS. Brilliant piece of software imo.
That's just from the uninstall????? Nothing else changed?
Nah i see what happened, you had a FPS cap to 60fps with MSi and the uninstall removed the cap That's my theory anyway.
Yes, I just uninstalled those 2, nothing else changed. I don't know about the FPS cap, I never set it manually. But it might have been the case. What was interesting for me before uninstalling these 2 is that I got good scores / FPS after a restart for a while (like a few minutes), then it went downhill and never came back until the next restart.
EDIT: Okay, I'm not entirely sure what's happening, it's pretty much back to where we started :/
Thats strange...try looking for a suspicious process in TaskManager while running the benchmark. Maybe MSI/Riva still running in the background.
Regarding the RAM, if u have a decent motherboard that can run 3200Mhz RAM (check HW forums if your MB can support that) then u can buy 2x8GB 3200MHz RAM for ~50-60$ - I recon it will give u 20-30% boost in this benchmark.
I took a look at a couple of tests and they show that, unless Raceroom is a memory-sensitive program, the difference between single-channel and dual-channel is usually very slim in real life usage (sometimes even to the point when single-channel gives marginally better results).
Its about single channel vs dual and 2400 vs 3200MHz -it should be significant.
see this example - older CPU with proper RAM (for that CPU) and result is 30% higher than Norbi930523
Anyway, he have other performance issues for now..
I don't think the difference is purely because of memory. He's obviously got some problems with the system, could be there were still some issues in the best run.
I have run the benchmark again, checked the Task Manager but didn't see anything suspicious (I also closed every background process that I could, but no difference). Also, the same processes were most probably running when I got the good benchmark scores. By the way, the Task Manager showed around 30% usage for CPU and around 98% usage for GPU.
As for the RAM, I have an MSI B150M Bazooka, which supports unfortunately only DDR4-2133MHz or 2400MHz with its DDR4 Boost if I understand correctly.
What's really strange for me is that I get (very) different results after restarts, even if I change nothing. I'm starting to think that a fresh Windows install would be worth a try.
The CPU usage looks correct as the benchmark uses ~1.5 threads and with your 4 cores that means around 30-40% CPU usage.
The GPU usage look strange - on my GTX1060 its on 50% during the benchmark. So it sounds like graphic card driver but you said you tried all the options in that area, so im out of ideas..
BTW, the GPU is still overclocked ?
What's your driver version by the way?
The GPU is not overclocked anymore (although I'm not sure why Unigine (or previously MSI Afterburner too) shows 2160MHz for the GPU clock, when according to MSI the Base clock for this GPU is 1500MHz and the Boost clock is 1830MHz)
I think your GPU clock is the default. Mine is also higher then stated on the specs (never OC it)...never understood it.
This is while running the Unigine. See the driver version there (didnt update it since ayear ago)
nvidia boost 3.0
So I tried a couple of things:
- get a 2x8 GB memory kit (at least I'm done with this upgrade too ): it's dual channel now, they are fitted in the slots where the motherboard manual recommends it, no change
- reinstalled Windows: this did fix 1-2 weird things that I noticed since fitting the new GPU, but unfortunately the FPS is not one of them I tested with a clean Windows right after installation, tested changing and benchmarking the settings one by one, also tested GPU drivers one by one
- I noticed that when I get low FPS in Unigine, the CPU usage is around 10% lower, in the 15-20% range in the Task Manager, therefore I disabled Intel TurboBoost in the BIOS, as there were some throttling-related settings to this feature, but didn't help either
So my FPS/scores in Unigine are still inconsistent, and I'm officially out of ideas I'm not sure what can cause the 10% CPU usage drop, but at least it shows that the suspect is around the CPU and not the GPU (at least judging from the fact that 15-20% CPU means low FPS/score, around 30% usage means good FPS/score).
Just out of curiosity, do you use your old monitor as a secondary display, or only the new 144 Hz one? And what happens if you scale the monitor's frequency back? Does it improve something maybe?
I only use the new monitor. After reinstalling Windows I left the frequency at 60Hz and started testing with that, then later changed it to 144Hz. I also tested Raceroom with both frequencies, it made no difference.
And what about cooling/temperatures? Are the operating CPU temps consistent through the "good" and "bad" benchmark results? But I see that GPU temps are ~45 °C which is fine.
Separate names with a comma.