Radeon RX 6600 Power, Temps, Noise, Etc.
The Radeon RX 6600 reference TDP is 132W, 28W less than the 6600 XT. AMD says the XFX card we received sticks to the reference spec, so let's see how it looks when we hook it up to our Powenetics testing equipment to measure in-line GPU power consumption and other aspects of the cards. We collect data while running Metro Exodus at 1440p ultra and FurMark stress test at 900p.
Power use during the Metro Exodus benchmark loop was just a bit more than the official TDP at 137W — nothing to worry about. Also notice how the card clamps down pretty hard at the power limit, whereas on the RX 6600 XT (an ASRock card) there was more available headroom. FurMark shows similar behavior, only with a slightly higher average power use of 140W. In both cases, the RX 6600 is a pretty big step down from the next lowest power GPU, the RTX 3060.
Clock speeds show a pretty big difference between FurMark and Metro, with the latter averaging 2485MHz while FurMark only averaged 2080MHz. AMD basically has two sets of behavior with FurMark clocks: The RX 6600 XT and RX 6700 XT run at around 2350MHz, and all the other RX 6000-series GPUs run at around 2080MHz. There's a lost more variation in clock speeds when running an actual game, with the RX 6600 hitting nearly 2.5GHz — it basically matches the boost clock of 2491MHz almost perfectly.
Despite having very different card and cooler designs, the fan speeds on the XFX RX 6600 are very similar to the ASRock RX 6600 XT. We don't see the nearly instant drop to 0dB mode between the Metro loops, but average RPMs are a bit lower than the larger 6600 XT card. In FurMark, the two cards are basically matched at 1340 RPM. That doesn't mean the cards behave the same at all, however. Fan speeds directly affect temperatures, and while the XFX card didn't run hot by any stretch, it was about 3–4 degrees Celsius hotter than the ASRock card.
Along with the Powenetics data, we also measure noise levels at 10cm using an SPL (sound pressure level) meter. It's aimed right at the GPU fans, in order to minimize the impact of other fans like those on the CPU cooler. The noise floor of our test environment and equipment measures 33 dB(A), and the XFX RX 6600 measured 39.4 dB while running Metro with a fan speed of 40%. I don't think the fan will normally go above 50%, but we also tested with a static fan speed of 75%, at which point the card generated 60.9 dB of noise and was quite loud. In normal use, however, it will likely stay closer to the 40 dB range.
Radeon RX 6600 Mining Performance
Ethereum mining continues to have an impact on GPU prices, and we ran some quick tests to see how the RX 6600 fares. We used NiceHashMiner to check mining performance with a variety of algorithms, though most of the tests failed to complete. The best option was Ethereum, not surprisingly, where after tuning we were only able to get about 28.5 MH/s.
The RX 6600 XT can reach about 32 MH/s, but it also comes equipped with 16Gbps memory, which can be overclocked to around 17.2Gbps. The RX 6600 in contrast comes with 14Gbps memory and we could only get that up to 15.12Gbps. That's about 88% of the memory bandwidth, after tuning, which dovetails nicely with the difference in hash rates.
That said, on that old i7, I think the only real problem would be possibly the motherboard. Sometimes, older motherboards are very finicky and won't work with newer video cards, most particularly when they're part of OEM systems like Dell and HP.
Anyway, thanks a lot for the review; much appreciated as always.
Wasn't the Z77 chipset PCIe2.0? It would make the 6600 siblings run in X8 of PCIe2, so I'd imagine they won't perform as in these charts? Maybe close, but I wonder how badly they'll be constrained.
EDIT: yep, 2.0 indeed: https://ark.intel.com/content/www/us/en/ark/products/64024/intel-z77-express-chipset.htmlEDIT2: Z77 with the 3770 does run in PCIe3.0; just want to clear that up just in case. It was pointed out later in the thread.
Luckily it's not a branded PC but a custom build.
In my case I would use the RX590 to replace my wife's HD7970 and get the RX6600 for myself IF I can get it at MSRP. If not, I will keep playing lottery at AMD's website in the hope of getting something realistically priced.
I still imagine it'll do far better than the RX 590, though.
As someone who is on a RX580 I think it would be better to go with the 6600XT model, but as other have suggested if you don't have a PCIe 4.0 motherboard makes this a more difficult decision.
Kind of the reverse of how a lot of 300/400-series AM4 motherboard could do PCIe 4.0 when using a Zen 2 CPU until AMD pushed an AGESA update to block it.
As for the review/RX6600 itself, basically feels like overpriced tech from years ago. MSRP is about $100 more than it would have any right to in a remotely sane economy.