The Intel UHD Graphics 620 (GT2) is an integrated graphics unit, which can be found in various ULV (Ultra Low Voltage) processors of the Kaby Lake Refresh generation (8th generation Core). I typically run stock clocks in games just because there isn't enough need for more performance on my 60 Hz monitors.Intel UHD Graphics 620 ► remove from comparison If I got a new 580 you can bet I wouldn't be worried about undervolting and would instead be putting as many volts through it as I could to get over 1500 MHz, at least in benches. Not when you have a central A/C unit running near constantly in south GA! I get that higher voltages lead to throttling and temp issues but its never been an issue with my main rig since its custom water and my backup uses an H80 which keeps the 2600K under 90C even at 1.52v and 4.8 GHz. Even massively overvolted PC's just don't use anywhere near enough power to be noticeable on the power bill in my experience. Also overvolt my CPU's as a matter of course (putting as much as 1.52v through my 2600K for 4.8 GHz, well, 4.7 GHz now since it finally degraded 100 MHz due to the harsh treatment its gotten all its life, and that chip is now over 6 years old). I overvolt the pee outta my Titans and always have (the only way to get them grudgingly over 1300 MHz). I've never given a hoot about power usage or efficiency personally. Not 1 month when they GPU is in it's prime state. They need stability in a wide variety of situations and they need to guarantee it for years. It is flawed methodology and why manufacturers don't do it. Using card specific voltage undervolting would inflate the the efficiency of any card*whether Nvidia or AMD), it is literally best case scenario(as it is just a hair over the crashing point) over a tiny sample range. This prevents the need to retest for stability each factory overclocked cards. Margins, particularly on AMD cards are not that big.Īdditionally with these manufacture overclocked cards, it is easier to overvolt the card. Doing the later is simply too costly and risky. It is simply more practical to give a average voltage which AMD likely gives them using statistical analysis which removes the guess work on optimizing a specific voltage for a specific GPU(not just polaris 10 family, every single rx580 would need to be tested and optimized). And not using a wide software selection is a guarantee for this. If it fails at the supplied voltage with the manufacturer, it is not working as intended and it requires a return. A game should be completely stable with the volts given and it doesn't matter the software used. If we was a GPU manufacturer, he would need to test through 20-30 games.īecause if that game crashes in just a single game because it doesn't have enough volts, an RMA for the card is needed and the profit from the videocard sale is erased. In addition undervolting to the bare minimum would take too much time(money) and lead to the GPU crashing in games sooner than later.Īdoredtv(flawed as usual) didn't test a wide variety of games. Everything else is seemingly identical save for the efficiency and clock speed gain due to switch in the process.Ĭlick to expand.Simply because you have to guarantee stability across a wide sample range and multiple game titles and for multiple years. Credit to them for catching Nvidia on one metric in which they've lacked in for years, but otherwise Polaris isn't all that impressive. If you look at any review it's quite obvious that Polaris isn't an efficiency winner against Pascal, but that's not too surprising given it only seemed to improve on three things over Fiji, polygon throughput, discard accelerator and colour compression. You can get either to be better by cherrypicking results from specific tests, but I don't really care about that. Summit Ridge also runs at a far higher voltage than Kaby Lake at a given clock speed but neither has a meaningful efficiency advantage in the real world. I'm not talking about power consumption as that's actually meaningful, but something like clock per voltage is as useless of a metric as pure clock speed. Click to expand.This is yet another case of numbers not being directly comparable.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |