GTX 970: 3DMark stated clock speed lower than what Afterburner is telling me

[pretiltedscales]

This is my first GPU, so please excuse my ignorance. I bought a used GTX 970 and I'm running benchmarks to be sure nothing is wrong with it. I d/l'ed and ran the free version of firestrike (3DMark). http://www.3dmark.com/3dm/17396804 shows that my core clock speed is 350 MHz. I thought this was weird as I believed it should be closer to 1050 MHz. I d/l'ed afterburner and ran firestrike again. Looking at the afterburner log, it tells me that my GPU's clock speed during the test would jump up to around 1225 MHz (1266 MHz max). I'm confused as this seems higher that it should be. What is going on?

 

[chemmajorp53]

Ignore 3dmark's readings, afterburner is more accurate, and I'd download GPU-z if you want to monitor your card more closely. The boost clock listed by the manufacturer is the max clock that your card is guaranteed to run at, not necessarily the clock that your card will top out at. If you have extra thermal and power limit headroom, then GPU Boost 2.0 will push your card past the advertised boost speed.

 

[pretiltedscales]

Ignore 3dmark's readings, afterburner is more accurate, and I'd download GPU-z if you want to monitor your card more closely. The boost clock listed by the manufacturer is the max clock that your card is guaranteed to run at, not necessarily the clock that your card will top out at. If you have extra thermal and power limit headroom, then GPU Boost 2.0 will push your card past the advertised boost speed.

Okay, so I shouldn't be worried about 3DMark's posted clock speed. I was wondering, b/c the overall score it was giving me was on par with other similar systems showing much higher clock speeds. Thanks!

 

[bystander]

If your result fell with in the normal score range, there isn't anything to worry about. They gather information about your GPU between the benchmarks, when your GPU may be at idle clock speeds.

 

[ErikMcLeod]

If your card was running at 350MHz, it'd probably take 3DMark an eon to wrap up a full benchmark, I'm talking barely double digit framerates and a pitiful score. Chances are, Afterburner is telling you the right clock speed, as 1200-1300MHz is about where GTX 970's tend to run without overclocking. Also, your score is just about where it should be, so there's nothing to worry about.

 

[Karadjgne]

My cpu is OC at 4.6GHz. I know this, bios knows this, cpu-z knows this. Every other program basically doesn't read actual speed, it just reads the cpu ID, stating the cpu is at 3.5GHz. This is entirely normal and will not affect actual performance. Does the same thing with my OC gpu, which is sitting at a 22% OC above factory settings. You will find most programs don't bother to delve into actual speeds, just read id's.

Take 3dMarks results with a grain of salt. While a decent gpu test, it's not only gpu that'll affect results. Cpu, mobo, ram, ssd can squiggle the results to one side or the other. It may be that even if ppl have higher clocks, they maybe using lower amounts of ram, so slowing info passage, might be running the test from a hdd or have a slower fx cpu than your cpu, so can't process the info as fast. In affect, they create a 'slow lane' and your gpu is in a faster lane respectively.

 

[bystander]

My cpu is OC at 4.6GHz. I know this, bios knows this, cpu-z knows this. Every other program basically doesn't read actual speed, it just reads the cpu ID, stating the cpu is at 3.5GHz. This is entirely normal and will not affect actual performance. Does the same thing with my OC gpu, which is sitting at a 22% OC above factory settings. You will find most programs don't bother to delve into actual speeds, just read id's.

Take 3dMarks results with a grain of salt. While a decent gpu test, it's not only gpu that'll affect results. Cpu, mobo, ram, ssd can squiggle the results to one side or the other. It may be that even if ppl have higher clocks, they maybe using lower amounts of ram, so slowing info passage, might be running the test from a hdd or have a slower fx cpu than your cpu, so can't process the info as fast. In affect, they create a 'slow lane' and your gpu is in a faster lane respectively.

As long as you look at the GPU score, or the details to see how your system is comparing to other similar systems, it's a great benchmark. The problem with so many other benchmarks, is you don't get results to compare to, and people rarely use the same settings. 3dmark is great about tracking everyone's results, making it easy to see if your system is having issues. It also tests both the CPU and GPU, which is important too.