HD 5870 first impressions

That's really curious thing - some put it being louder than HD4870X2 and what not, but then other sites are saying it's the most silent reference cooler ever

 

My impression is that HD5870 is less noticeable than HD4870 I had before.

 

Easily explained.

Those that like to "play" furmark report high noise levels.

Those that actually test in real applications and games report low noise levels. With regards to mine, it's quieter in games than either my 4870/512 or 4890. At least it's less noticeable. My PSU right now is the biggest noise generator, and it's a relatively quiet PSU. But going to replace it with a quieter one. It's over 4 years old now after all.

And finally sold my 4870. Heh, glad people still like getting deals on older hardware.

Regards,
SB

 

I have been playing Risen for hours, and the fan stays around 30%.

Edit: that with the GPU overclocked to 900

 

It's less noticeable than my 4870 or 4890.

 

Has anyone of you guys, being using his Evergreen with a HDTV?

 

Not yet! First I must put my computer back to new bigger case

Going back to card loudness I've noticed that some models are using 1.125V for GPU under load (like mine Shappire) and other are using 1.165V under load. This might explain quite big variations between different reviews in power consumption tests under load ...

 

Actually, it does the opposite! We scale the voltage based on leakage, so the higher leakage parts use lower voltage and the lower leakage parts use a higher voltage - what this is does narrow the entire TDP range of the product.

Everything is qualified at worst case anyway; all the TDP calcs and the fan settings are completed on the wors case for the product range.

 

So the better parts are artificially made to draw more power?

 

I guess so, but they would overclock higher.

 

No, the nominal voltage is 1.165V, however the higher leakage parts do not need as much voltage to run at those speeds.

 

I'm not sure lower leakage means better part. If the lower leakage is due to thicker gate insulator, then the transistors might switch slower, and they'll actually need higher voltage to reach the same speed.

 

What Dave is trying to say is that higher leakage parts needs less volts to reach target freq. In other words higher leakage = higher clocks at a cost of higher power consumption. It's like with AMD Phenom II TWKR edition, it is a very high leakage part, but it takes cold well and clocks exceptionally!

So I'm paying a bit more from electricity thanks to my card, but I'm hitting 1GHz core FurMark stable at only 1.18V

Lower leakage wastes less electricity under same conditions (like 850MHz @1V) and also in power saving mode (but PP gating is so good on RV8xx that it probably makes really small difference anyway).


BTW thank you Dave for clearing that up!

 

I've been playing with a 5870 as well and it's quite an impressive card. In addition to just being really fast in normal rendering, the compute shader implementation seems pretty solid already. It's also a really nice way to develop graphics stuff compared to CUDA, etc. IMHO - no (horribly bad) interop penalty, no need to set up additional resource types or copy data, etc. It all just works with the graphics stuff you already have set up.

 

OK I guess that makes sense. Thanks for clarifying that!

 

Higher leakage part have thinner insulator layers ... which can run at reduced voltage while meeting the same timing requirements. Now obviously this isn't going to get you back to exactly the same power consumptions with the same timing margins, but you can claw some of the losses back.

 

Lightman is 100% right the reason the high leakage part has a lower voltage is because they don't need as high a voltage to reach the same clock speed.