Were Mistakes Made? Radeon RX 7900 GRE Review Update

[Music] welcome back to Harbor onbox today we're taking another look at the 7900 gr and this is a re revie of the re-review of the review and I'm taking a third crack at reviewing the jar because there was a small amount of controversy regarding our results which was a result of us opting to do what we always do and that is test gpus at their base spec using reference or found Edition models when possible and when not possible we manually clock them to the Bas spec or load a Bas speec bios the problem with this Approach at least in this instance is that the AMD reference version of the gr it runs at a lower power Target than many of the board partner models or perhaps all of them it also runs much hotter and therefore could be prone of throttling we have seen this from AMD reference cards in the past but given that we always use base model data for our GPU testing and we did so when testing the 7800 XT for example it didn't seem right to compare something like the sapphire nro plus with the AMD reference 7800 XT and admittedly I really didn't think it'd be that much of an issue anyway because normally board partner models are just a few per faster but when comparing the AMD reference 7800 XT with the reference 7900 gr we found the gr was just 6% faster in average for rization performance and 10% faster when focusing on R tra ing now if we compare those numbers with a trusted meteor Outlet such as Tech PowerUp who do excellent testing we see they found the sapphire Nitro plus version of the GRE to be 133% faster than amd's reference version of the 7800 XT at 1440p and then 12% at 4K so around twice the margin I found that said they also found the sapphire Nitro plus version of the 7800 XT was 4 to 5% faster on average which would push the margin more in line with what I reported so I've gone back and added the sapphire Nitro plus versions of the 7800 XT and 7900 gr to my data something I wish I had have done originally but didn't realize at the time that this would be a potential issue this will allow us to not only compare overclocked versions of both models but also cross reference that data with the AMD reference models and there's some really interesting data in here that highlights just how unusual the gr can be so let's get into it starting with Resident Evil 4 at 1440p the Nitro plus jar is 7% FAS than the AMD model so that's a big gain there meanwhile the Nitro plus version of the 7800 XT is just 3% faster than the AMD reference 7800 XT so when comparing AMD reference models the gr was 8% faster but if we compare the Nitro plus models the jar is 11% faster and worse still if we compare the Nitro plus jar to the Bas speec 7800 XT it's 15% faster and that's the comparison that we were originally trying to avoid and it's a similar story at 4K the Nitro plus 7800 XT is just 3% faster than the reference model while the Nitro plus gr is 5% faster so whereas the AMD reference model saw the G win by a 9% margin using the Nitro plus sees it win by a 12% margin again if we compare the stock 7800 XT to the Nitro plus J the Sapphire card is 15% faster now this is odd though retesting with a plagu toil requium sees no performance change for either of the Nitro plus models performance is the same as the AMD reference cards at 1440p and even at 4K we find the same thing the overclocked Nitro plus models delivered the same level of performance that said though this isn't necessarily unheard of we have seen this in the past from Factory models they don't always deliver more performance and it's a similar story in cyberp Punk 2077 Phantom Liberty the Nitro plus models deliver similar performance to that of the AMD reference models and this was seen at both 1440p and 4K d light 2 though does see some performance uplift with the Nitro plus models at 1440p in the case of the 7800 XT we're looking at a rather large 6% boost with just a 2% boost for the gr which is a bit surprising now increasing the resolution to 4K does eliminate the gains from overclocking the 7800 XT and now the Nitro plus models are just 2% faster so like what we saw with Cy Punk and a plague tail requium there's really not much in it next up we have Spider-Man remaster and here the results do favor the Nitro plus jar which is 8% faster than the AMD reference version at 1440p whereas the Nitro plus version of the 7800 XT is just 1% faster than the reference version that means if we compare the AMD reference spec versions of the 7800 XT and GRE the GRE is just 5% faster but if we compare the Nitro plus models the jar is 12% faster and if we compare the base spec 7800 XT to the Nitro plus J the margin grows to 14% And it's a similar story at 4K here when comparing the reference models the jar is 8% faster but when comparing the Nitro plus models it's 11 % faster moving on to Hogwarts Legacy at 1440p we see that when comparing the AMD reference models the gr is 9% faster whereas that margin grew quite substantially with the Nitro plus models to 177% and that's a 23% uplift if we compare to the base spec 7800 XT then at 4K as we saw previously a 133% increase for the gr When comparing amd's reference models and with the Nitro plus models that margin increased to 17% so if you were to compare the Nitro plus jar to a base model 7800 XT you'd find it to be 22% faster in this Example The Last of Us Part One results are surprising in that they're similar to what we've seen in a few other titles such as cyber Punk so that is to say little to no difference between the jar and 7800 XT regardless of the model used we know this to be a very memory sensitive title so it would seem as though the more memory Limited bandwidth of the gr is a problem here and it's much the same at 4K that said we do go from a 2% margin in favor of the gr when using the reference models to 4% with the Nitro plus now next up we have Starfield and here the margins do increase when comparing the Nitro plus models previously we found just a 2% margin when using the AMD reference versions but with the Nitro plus cards that margin grows to 8% still not significant but much bigger than what we previously saw then at 4K we go from no performance difference between the 7800 XT and J with the reference models to 10% When comparing the Nitro plus cards testing with Call Of Duty Modern Warfare 3 shows no real change between the Nitro plus and the reference models we're still seeing a 15% uplift for the gr at 1440p which is quite a substantial Improvement at least based on some of the games we've seen in some of the other games tested and that margin is then extended to 21% 4K so this title seems to scale really well for the G now with Alan Wake 2 we previously found when testing the base spec models that the GRE was just 3% faster than the 7800 XT at 1440p but that margin grew to 8% when using the Nitro plus cards and it's a similar story at 4K previously the gr was just 3% faster but when using the Nitro plus it's now 6% faster and if we compare the Nitro plus jar to the stock 7800 XT we see a 9% increase the gr also looks more impressive in watchdog's Legion when testing with the Nitro plus previously the J was found to be just 4% faster at 1440p but when comparing the Nitro plus models that margin blows out at 10% then at 4K we find no difference in performance between the 7800 XT and GRE but with the Nitro plus cards the gr is now 6% faster finally we have Hitman 3 and here the margins don't really change at 440p the gr was just 3% faster when comparing the reference models and that margin is actually reduced to just 2% with the Nitro plus cards that said if you were to compare the stock 7800 XT to the Nitro plus gr the gr would end up being 6% faster the 4K results though they're quite different when comparing the AMD reference models the G was 10% faster so quite a large margin and it's only slightly extended we're comparing the Nitro plus models to 13% but again if we compare the stock 7800 XT to the Nitro plus GRE we do find a 177% margin in favor of the J okay so here's the 12 game average focusing on the 1440p data previously we found the 7900 GRE to be just 6% faster than the 7800 XT and if we compare the Nitro plus models that margin only increases to 8% however if we compare the base model 7800 X2 to the Nitro plus GRE we find an 11% margin which is very close to the 133% margin that tech power up found in their Nitro plus jury review using a range of different games so that data makes sense now at the 4K resolution we previously found the G to be just 5% faster than the 7800 XT and when comparing the Nitro plus models that margin grows to 11% which is the same margin we see when comparing the base model 7800 XT to the Nitro plus gr so the same margin seen at 1440p and please keep in mind these average results are calculated using the geom now if we were to recalculate the cost per Frame data based on the 1440p results we find much the same the 7800 XT is slightly better value but overall they're much the same in terms of value across the Dozen games that we tested even if we compare the Nitro plus GRE results with the base model 7800 XT in terms of value they're much the same also just in case you're wondering I have put question marks next to the Nitro plus models as they're not really $500 US MSRP models but for the sake of this comparison let's just pretend that they are the gr does shape up better at 4K when compared to the Nitro plus as both these 7800 XT and gr end up offering about the same value but really there aren't any significant changes here so there you have it I'm not going to bother going over all the ray tracing data again I think you get the idea and that is if you compare like for like the margins really shouldn't be that dissimilar overall to what we've shown previously for example the Bas speec models saw a 6% advantage in favor of the gr whereas the overclocked Nitro plus models saw an 8% margin so like for like the 7900 gr really isn't that much faster than the 7800 XT across a wide enough range of games there are certainly examples where that 33% increase in cause results in around a 20% performance boost but there are many more examples where the gains are 8% or less again these results make sense with what others such as Tech PowerUp have reported they've shown for example that the Nitro plus 7800 XT is 45% faster than the reference 7800 XT and in fact it matched the 6900 XT exactly in those tests so if you were to take that as a reference point you'd find that the 7900 J Nitro plus is 8% % faster than the 6900 XT which aligns exactly with our data so again overall the 7900 GRE it's not world's faster than the 7800 XT I've seen people in the comments trying to tell me it's 15 or 20% faster but that's simply not true if you test enough games rather those margins are up to or best case scenarios they're certainly not what you can expect to see across a wide range of games as for how the 7 900 gr should be compared to the 7800 XT it's difficult to say as partner card models are still yet to go on sale or at least they were the time I Was preparing this video ideally though I will need to get my hands on a $550 MSRP model and test that but I think it's fair to say that as long as you're not comparing a premium model such as the Nitro plus to a base model 7800 XT and doing your value analysis based on that then I don't really think there's an issue at the end of the day none of this changes our recommendations or sort of lack of enthusiasm for the gr like if you're really excited about the GRE then you are probably really excited about the 7800 XT which you know wasn't a terrible product in terms of value but if you're really excited about the 7800 XT then you should have been really excited about the 6800 XT which was a good product admittedly you just couldn't get it during the mining boom craze but the point is we're many years on from the 6800 XT which we believe was a good product uh and these newer rdna A3 models while they do improve the value it is many years later not that exciting but what is a bit exciting is that amd's told me that the extremely limited overclocking of the GRE cards is a bug and it will be addressed shortly hell it might even be fixed by the time you watch this and if that is the case I will be Keen to do some overclocking to see just how much more you can squeeze out of cards like the Nitro plus we're not limited to a 3% memory overclock for now though we are done here if you enjoyed this sort of follow-up content where we tried to explain the differences you were seeing in reviews and the difference between a product like this and a base spec at 7800 XT which I'm not convinced is a good way of evaluating the GRE but again we'll have to get some sort of Base speec MSRP models of the jar and see how they differ to this probably not going to be a big difference because there's not really a big difference anyway but it does skew the results a bit using a premium product like this against a reference 7800 XT anyway like if you liked it subscribe for more content and if you'd like to uh support the channel and get some cool perks and return we have float plan or patreon signing up to either one of those gives you access to stuff like our Discord chat straight away members were asking me shortly after the reviews went live about the discrepancies between various different review outlets and what was going on there and I was able to do this testing fairly quickly and inform them as I was doing uh the testing so on a game by game by game basis I was able to provide results and let you guys in the Discord chat know what I was Finding and what I thought was going on so yeah interesting for those of you who want a bit more insight we also do monthly live streams behind the scenes content and Q&A stuff so check that out if you're interested but if not that is perfectly fine and I would like to thank you for watching this video I'm your host Steve and I'll see you again next [Music] time [Music]

Channel: Hardware Unboxed

Views: 149,609

Keywords: hardware unboxed

Id: lDp9rT4YUpA

Length: 15min 39sec (939 seconds)

Published: Wed Feb 28 2024