- Mar 16, 2014
- 183
- 0
- 10,760
Evening.
Recently, my mate bought a brand new Nvidia GeForce GTX 970. The ASUS Strix model (STRIX-GTX970-DC2OC-4GD5) specifically. He upgraded from an Nvidia GeForce GTX 660 DirectCUII (GTX660-DC2O-2GD5), expecting a major increase in performance.
His system consists of an Intel Core i7-4820K (standard core clocks and voltages), his brand new Nvidia GeForce GTX 970 Strix, Corsair Value RAM (one stick, 8GB, 1333Mhz, CL9), ASUS P9X79 Motherboard (BIOS version 4502) and a Corsair CX750 PSU (...which I'm a bit unsure about on the quality side).
However, immideately, he noticed that games seems to run very sluggish and not perform at all up to the task expected. I popped over to him, and we tested the system together. I ran some benchmarks, and the results were not good. I even compared the tests to my own system, with an Intel Core i7-4820K (overclocked to 4200Mhz), an Nvidia GeForce GTX 760OC DirectCUII, HyperX Savage RAM (two sticks, 16GB, CL9), and an EVGA G2 750W PSU.
When I ran the Benchmark, I first ran Valley Benchmark alone. These are the results:
The first run looked great. Decent score and with a stable, 100% usage on the GPU core. Compared, the score was a lot higher than my setup. However, on the next run, when we ran it with Prime95, to push the CPU-cores to 100%, with the Blend test, the score was a lot worse.
This is the Valley Benchmark score with Prime95.
Obviously, pushing the CPU to 100% together with the Valley Bechmark-test, will reduce the performance. But, on this very test, even my system performed a lot better with the Nvidia GeForce GTX 760OC. We noticed that GPU usage never reaced over 35%. In my mind, this is caused by the CPU not beeing strong enough to recieve all the data from the GPU. Bottlenecking, effectively. In all Synthetic tests in Valley Benchmark, I used the Extreme HD-preset.
Temperatures were great during all the tests. But, since the GPU never got higher than 35% in the last test, the temperature never rose. Neither did I notice that his CPU core clock boosted from 3700Mhz to 3900Mhz. I confirmed in BIOS that Turbo Boost was indeed enabled. He told me that the Drivers was up-to-date, however I updated the Chipset Drivers anyways. He also told me that he had some issues with the Graphics Drivers, though, so he rolled them back to a previous version. Nothing of what we did solved the issue, though.
Neither are these problems only associated with Synthetic Benchmarking, but he has told me that games struggle to maintain a higher FPS and occational stuttering is occuring, occationally (lots of occations, there).
What can cause this issue? As the i7-4820K is a professional CPU in the X79 platform, designed for SLI-configurations, how come it will cause situations comparable to Bottlenecking with a single Nvidia GeForce GTX 970? Does Maxwell have issues with X79? Is the cheap CX-PSU causing problems, or, is it the rolled-back driver that's causing issues?
This one really messed up my mind.
Any help appreciated, to make this bloke happy.
Thanks!
Recently, my mate bought a brand new Nvidia GeForce GTX 970. The ASUS Strix model (STRIX-GTX970-DC2OC-4GD5) specifically. He upgraded from an Nvidia GeForce GTX 660 DirectCUII (GTX660-DC2O-2GD5), expecting a major increase in performance.
His system consists of an Intel Core i7-4820K (standard core clocks and voltages), his brand new Nvidia GeForce GTX 970 Strix, Corsair Value RAM (one stick, 8GB, 1333Mhz, CL9), ASUS P9X79 Motherboard (BIOS version 4502) and a Corsair CX750 PSU (...which I'm a bit unsure about on the quality side).
However, immideately, he noticed that games seems to run very sluggish and not perform at all up to the task expected. I popped over to him, and we tested the system together. I ran some benchmarks, and the results were not good. I even compared the tests to my own system, with an Intel Core i7-4820K (overclocked to 4200Mhz), an Nvidia GeForce GTX 760OC DirectCUII, HyperX Savage RAM (two sticks, 16GB, CL9), and an EVGA G2 750W PSU.
When I ran the Benchmark, I first ran Valley Benchmark alone. These are the results:
The first run looked great. Decent score and with a stable, 100% usage on the GPU core. Compared, the score was a lot higher than my setup. However, on the next run, when we ran it with Prime95, to push the CPU-cores to 100%, with the Blend test, the score was a lot worse.
This is the Valley Benchmark score with Prime95.
Obviously, pushing the CPU to 100% together with the Valley Bechmark-test, will reduce the performance. But, on this very test, even my system performed a lot better with the Nvidia GeForce GTX 760OC. We noticed that GPU usage never reaced over 35%. In my mind, this is caused by the CPU not beeing strong enough to recieve all the data from the GPU. Bottlenecking, effectively. In all Synthetic tests in Valley Benchmark, I used the Extreme HD-preset.
Temperatures were great during all the tests. But, since the GPU never got higher than 35% in the last test, the temperature never rose. Neither did I notice that his CPU core clock boosted from 3700Mhz to 3900Mhz. I confirmed in BIOS that Turbo Boost was indeed enabled. He told me that the Drivers was up-to-date, however I updated the Chipset Drivers anyways. He also told me that he had some issues with the Graphics Drivers, though, so he rolled them back to a previous version. Nothing of what we did solved the issue, though.
Neither are these problems only associated with Synthetic Benchmarking, but he has told me that games struggle to maintain a higher FPS and occational stuttering is occuring, occationally (lots of occations, there).
What can cause this issue? As the i7-4820K is a professional CPU in the X79 platform, designed for SLI-configurations, how come it will cause situations comparable to Bottlenecking with a single Nvidia GeForce GTX 970? Does Maxwell have issues with X79? Is the cheap CX-PSU causing problems, or, is it the rolled-back driver that's causing issues?
This one really messed up my mind.
Any help appreciated, to make this bloke happy.
Thanks!
Facebook
Twitter
Instagram