A Replacement for GPU-Z
Posted 24 January 2014 - 05:16 AM
This can allow you to include CPU usage and max single thread usage to see whether or not you are bottlenecking your GPU. Not only that, but they have announced that in their next update they'll be including a hotkey option for when to start the logging which will allow us to hit F11 and run both FRAPS and HWINFO at the same time. It saves in .txt as well and also in a format that can be easily imported to spreadsheets.
That's an image of what I have it set to show, it actually has an amazing list of different usage that you can monitor and is easy to edit to show/log what you want. It can also display the minimum, maximum, and average next to the current but it is all optional. Another setting that can change is it's refresh rate (default is 2000ms, so you will want to change it to 1000ms to correlate with FRAPS).
Just a thought since having the hotkey would be the big one to help benchmark. I have compared it with GPU-Z and they both seem to show almost the same results (there are very little differences in their readings) Here's the link in case anyone would like to test it for themselves: http://www.hwinfo.com/
Oh and I forgot to mention, it's ad-less free-ware!
Posted 24 January 2014 - 06:05 PM
Posted 26 January 2014 - 02:37 AM
Posted 26 January 2014 - 07:20 PM
Posted 04 June 2014 - 07:15 PM
Does it correctly report dedicated video memory usage under Windows 8? From what I have seen and read, GPU-Z does not.
Posted 04 June 2014 - 11:36 PM
HWInfo is what you should be using. CPU-Z almost never report the right stuff and I have to assume there is something wrong with GPU-Z as well. It's tied for the most comprehensive list of sensor readings I've seen outside of Everest and you have to pay for that one, no brainer really especially since it shows you all of the different temperature sensors which for Haswell is a godsend.
Posted 05 June 2014 - 09:13 PM
My current set of benchmarks is using GPU-Z (I would have switched earlier had someone told me to in my thread... and I did ask), so I'm going to finish out with that. But if I ever decide to do this again, I'll use HWInfo, instead.
Posted 08 July 2014 - 01:38 PM
If I should be using hwinfo instead of gpu-z, is there a guide on it somewhere? There's a lot of options and things to track and I'm not sure which ones are relevant.
Posted 08 July 2014 - 01:43 PM
I use HWInfo to see both CPU and GPU usage ingame using RTSS anyway, sounds like a good option to benchmark.
Posted 08 July 2014 - 03:50 PM
There's like eighty things it tracks. Many of which are named with abbreviations I'm not familiar with. The choices in the Help menu are "About" and a forum link. I don't see anywhere a basic "getting started" sort of guide. I don't suppose anyone else knows where one is?
Posted 08 July 2014 - 04:06 PM
It's a bit tough if you aren't much of an overclocker/enthusiasist. I don't understand more than half of the stuff either.
The worst thing is some of the sensors change depending on the hardware you have installed, so it's pretty much impossible to give any tips.
Personally, I just watch the CPU temperatures across individual cores, fan speeds, and the first set of values for graphic card, which I assume are related to the GPU in most cases. For me it shows the basic stuff GPU-Z did: temperature, clocks, memory utilization ... I hid most everything else :P (but the activity monitoring for disks is pretty nice).
Edited by Octopuss, 08 July 2014 - 04:08 PM.
Posted 26 July 2014 - 04:43 PM
I second HWiNFO, and note that it includes support for the RivaTuner Statistics Server, which means FRAPS is really unnecessary; if you want to record too, just use MSI Afterburner, which works on all GPUs (and even has hardware encoding acceleration for Intel, Nvidia, and AMD).
(In fact, for recording, if you have a modern Intel processor with an integrated GPU, you can use QuickSync to encode high-quality H.264 at virtually no performance cost.)
And when using HWiNFO with modern GPUs, some things to look for:
- Modern GPUs have aggressive power-saving features, including "Turbo", where they won't be at the same clock speed all the time. Therefore, since HWiNFO can record to a CSV file, it may be useful to check out how long your GPU spent at clock speeds below maximum when your GPU usage was well above 0. High-end modern GPUs especially will aggressively downclock based on thermal limits. For instance, the R9 290X with a stock cooler will not stay at its default max clock of 1000MHz for long, since it is intended to run only ~5-7 degrees below its thermal limit of 95C when near max load. (Non-stock cooling seriously helps here. )
You can use the same method to find out if your processor is being thermally throttled, too.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users