AMD Radeon RX 5500 (NAVI14) gets benchmarked and does hold ground

Published by

Click here to post a comment for AMD Radeon RX 5500 (NAVI14) gets benchmarked and does hold ground on our message forum
https://forums.guru3d.com/data/avatars/m/175/175902.jpg
Nice news, too bad it is still a X8 (as RX 460 550 and 560), the price is OK, but for the Black friday sale you can have a RX 590 8go for 15 Euro more... Waiting a review of the not OEM version in here 🙂
https://forums.guru3d.com/data/avatars/m/248/248994.jpg
Entry level video cards are starting to beat my dusty old GPU...
https://forums.guru3d.com/data/avatars/m/198/198862.jpg
Its true when they said it performs the same as rx480. Not bad for a 150$.
https://forums.guru3d.com/data/avatars/m/250/250418.jpg
I'm a bit disappointed, had hopped it would marginally beat the RX590. But what disappoints me even more, is the 7th October lunch and I can't buy it.
https://forums.guru3d.com/data/avatars/m/220/220188.jpg
gtx1660 is still faster and somehow more power efficient even though its on 12nm vs 7nm, they better sell for 120$ or something
data/avatar/default/avatar27.webp
EspHack:

gtx1660 is still faster and somehow more power efficient even though its on 12nm vs 7nm, they better sell for 120$ or something
And...?
https://forums.guru3d.com/data/avatars/m/246/246171.jpg
Vlooi:

And...?
What else are you expecting? EspHack made a complete coherent point - the RX5500 needs to be about $120 in order to be appealing considering the competition. I don't see what else needs to be said about that.
https://forums.guru3d.com/data/avatars/m/273/273678.jpg
EspHack:

gtx1660 is still faster and somehow more power efficient even though its on 12nm vs 7nm, they better sell for 120$ or something
the 5500 is a 1650 competitor, while the power util looks worse than the 1660's, it might undervolt to(or below) the 1650, AMD just has a habbit of letting their cards have more voltage than really needed.
https://forums.guru3d.com/data/avatars/m/209/209146.jpg
Some binning would be nice but I suppose that'd be pretty silly marketing some high-end low availability high efficiency edition of GPU's and variance is still big enough to where some cards just don't undervolt though that could have been solved by a less extreme clock target. (Would also lower temps a bit not going for those last ~2% gains for 30% additional power draw total and a 10 Celsius junction temp increase for example.) Some undervolting of the 5700 and XT models are actually used by almost all bios files though for the SOC (System on Chip) with a voltage value targeting 1050 instead of 1200 which is used by the stock and Anniversary editions from AMD themselves matching the 1.2v core voltage. (Need to look into the review itself I'd hope the 5500 here would be a bit less than that at least. 😀 ) Though this has kept up since at least the Fury as has the binning and variance so some can do -50 others almost -200 on voltage and others just don't undervolt so short of AMD having some more room for binning and taking a hit on the chips that would pass that's not looking like it's going to change. Makes it interesting for what the Navi that scales upwards is going to do though and how much headroom there's here not that it affects the product or reviews and I expect it to be the same as usual and testing is done on stock values after all. Otherwise with the OC and voltage headroom the 2080 or 2080Ti could also be OC'd and likely would win against the rumored 5800 or depending on how things go maintain the lead with a bit more of a gap. That 2080 and the Ti variant card has been out for a while not so "merely" catching up with bigger Navi isn't a huge win for AMD either and a small lead could be countered by if the higher end cards use HBM simple price/performance ratio and how AMD might not have much adjustment room here either and the upcoming 3000 series from NVIDIA for I guess the later half of 2020 for the general consumer market cards according to current rumors I think? (Just how I see it and even if the 5900 can pull ahead by a bit it might not be very impressive in the long run particularly if NVIDIA is readying or has already presented their next GPU models when AMD's Navi12's or what they'll be hit retail.) EDIT: Well I might be a bit biased to AMD as a long term user of their GPU's but even with that it's kinda hard not to notice the kinda skewed results from a minimal performance drop and then just how much of a core voltage reduction this could allow for and the resulting drop in power draw and heat thus also making the card cooler and the fans more silent for like 2 maybe 4% or so performance loss if it registers at all. (And the bigger drops are only really going to be in specific benchmark tests not that getting the lead high score here isn't a big thing but it is kinda extreme.) Though I do not find the power draw or voltage itself too problematic it's within what the card can handle after all but compared to the competition it is a factor and then there's the results for how much it affects GPU junction and possibly memory junction temps and other components and what can be done both for a manual fan curve and the automatic default one. (Even if the current drivers seem a bit hit & more than a bit miss for the readout here currently but that's a separate issue hah. 😛 ) EDIT: Maybe something for the refined 7+ nm and Navi 20 though of course the chances are AMD could just push those even higher with the improvements provided from whatever quality the resulting chips and yield ends up being here alternatively keep it on the same level and having a bigger batch of chips if it's the way I'm thinking this works on so more of these could pass and be fine. (Allowing a bigger volume and better availability of cards as a result.)
https://forums.guru3d.com/data/avatars/m/237/237771.jpg
This is about 20% faster than the 1650 but uses twice the power. A bit underwhelming at $150 it's not a bad choice but consider the 1650 can do this without a PCI 6pin power supplement. I"m more curious about the 5600's and 5800's cards. And to those talking of undervolting. NEVER BASE A DECISION ON THIS, IT'S NOT GUARANTEED TO UNDERVOLT JUST LIKE IT'S NOT GUARANTEED TO OVERCLOCK. Always base your decision off of out-of-the-box performance.
https://forums.guru3d.com/data/avatars/m/246/246171.jpg
Loophole35:

And to those talking of undervolting. NEVER BASE A DECISION ON THIS, IT'S NOT GUARANTEED TO UNDERVOLT JUST LIKE IT'S NOT GUARANTEED TO OVERCLOCK. Always base your decision off of out-of-the-box performance.
I'd rather say "always based your decision based on whether the product will fulfill your needs at the desired price point". Some games/programs run better on Nvidia, some run better on AMD. If you care about wattage and if this is known to be good at undervolting, then that's an ok reason to decide on it, as long as you're aware that your mileage may vary.
data/avatar/default/avatar03.webp
Is there going to be a 5600?
data/avatar/default/avatar08.webp
Yes. the 5600XT is rumored to be released in January 2020. US 249 is the expected price. It will have 192 bit mem bus resulting in 336 GB/s (5500 has 224GB/s) bandwith. It is expected to be in the 1660 TI to 1660 super range. I hope it will even be a little bit faster 🙂
https://forums.guru3d.com/data/avatars/m/189/189980.jpg
Let's see them benchmarks.