[00:00:00] What if we take a 7800X3D and put it up against a 7950X3D? 8 cores vs. 16 cores, 16 threads vs. 32 threads? Oh my god, the 7950X has as many cores as the other one has threads, this will be brutal.
[00:00:16] We already did a 7700X vs. 7800X3D and a 7950X vs. 7950X3D comparison. Both of which made perfect sense. Both same level comparisons had an explainable bump in price for a relative bump in performance. But what if we take the lowest tier Zen 4 X3D chip and compare it directly against the absolute highest Zen 4 X3D chip? Am I bored and trying to put numbers onto something that’s already obvious or does this comparison actually make sense?
[00:00:53] Of course I have a reason for doing this, why would I make a video about something obvious? Starting off with power consumption and temperature. Yeah, roughly half. At 143W running Cinebench All Core, the 7800X3D clearly wins with its seemingly tiny 84W. And the same then translates into cooling. Although the 7950X3D isn’t that hard to cool down, the 7800X3D is so much easier to cool down that it will actually work in a well equipped small form factor build.
[00:01:28] And now, before we get to the point of this whole video, let’s get to the brutal part. At full blast, the 7800X3D can manage up to 5.05GHz on a single core and the 7950X3D can do 5.65. On all cores, the 7800X3D can maintain 4.825GHz and the 7950X3D can do 4.8 on half of them and 4.9 on the other half. But the problem is that the core count for the 7950X3D is double that of the 7800X3D, so you can see where this is going.
[00:02:00] For every benchmark we did, the 7950X3D just rolled over the 7800X3D. And of course it did. Double the core count and up the speed on the same architecture and it’s going to be a bloodbath. Blender monster. Junk room. Classroom. Double the results in each and one of them. 3DMark CPU profile test now nicely shows step by step how this is happening. Starting off at 14% more, the two were plus minus equal until the core count of the 7800X3D just gave up and the 7950X3D dominated at 79% plus. PC more, 10% more. CPU-Z, 106% more in multi and only 6 in single. Cinebench, 12 more on the single and double on all core. Passmark, 87% more. Handbrake, it took 46% less time on the 7950X3D than the 7800X3D. And as a last good metric, double on the new Corona 10 benchmark.
[00:03:02] All of this was definitely foreseeable. There was no way it would have ended any different. As long as it has anything to do where core count plays any role or core speed or the two play the majority of the roles, this is just what we’re going to see. And we actually averaged that out. Going from a 7800X3D to a 7950X3D gave us about 57% more performance on average across all the benchmarks I just mentioned. And that’s just because this includes the tiny single core performance stats like CPU-Z single or Cinebench single. Or for example the low thread count 3DMark CPU profile runs. If we were to take these out of the average, it’s much closer to a double.
[00:03:49] Having that out of the way, let’s get to the point that I’m trying to make. Using the same GPU, there isn’t really a win in performance in games. On CSGO, we saw a very small bump in everything except for the 0.1% lows where the 7800X3D took the win. On Shadow of the Tomb Raider, it switched. From average to 1% lows, the 7800X3D won. Far Cry 5, the same thing. Rainbow Six Siege, again, win for the 7800X3D except for the 0.1% lows. And the only game that really proved to work better on a 7950X3D is Dota 2 where every metric won.
[00:04:34] And we averaged these numbers too, of course while taking out the 0.1% lows because if my neighbor sneezes, it’s a 100% difference. So averaging the min, max, 1% and average FPS, the jump from a 7800X3D to a 7950X3D is actually minus 2%. So you lose performance. And this now paints a totally different picture for me. We all knew that more cores does not equal more battery in games, but the 3D chips are specifically designed for gaming. Or at least the only game you can really observe in a real-world scenario is during gaming. And going from a 7800X3D to a 7950X3D does not give you any benefit at all. And sometimes it even loses.
[00:05:27] In the few games that we ran, at some point this became a loss, but if you add more and more, it will equalize at some point, maybe become plus 2% if you add the right games, but there is no win. So if you look at what use case pairs well with what chip, for the 7800X3D, it’s gaming. That thing is a pure gaming. It beats pretty much everything out there right now and it even competes with the top of the line chips on the same level. That said, production work-wise, it can’t keep up with the theoretically slightly lower tier 7700X. So if it’s a raw work PC that you are building, the 7800X3D is just not a good fit at all. At least compared to something like a 7950X, of course.
[00:06:16] Now for a 7950X3D, it’s a bit the best of both worlds. It’s almost as good as a 7950X in working, and it’s almost or equally good for gaming as a 7800X3D, and it’s a lot better in gaming than a 7950X. So with minimalistic drawbacks, it has the best of both worlds. And that’s kind of the conclusion here. If you are sitting on the decision between a 7800X3D and a 7950X3D, look at what you are doing. Are you really pushing these cores using software that you can make use of it? And are you gaming a lot? Then sure, go for the 7950X3D. You will get both. However, if you are building a PC that’s only for gaming, the 7950X just does not make sense. This is not going to be a, let’s just take the best one and we will have the highest FPS counter situation. You will lose, maybe just slightly, or you will win a percent, but compared to the 7800X3D, you won’t win. And it will be a lot harder to cool, and it will consume a lot more.
[00:07:20] On the other hand, if it’s a pure working PC, neither of them will work fine. The reduced clock speed and 3D cache just doesn’t help you in any software, and for these cases, just skip the 3D line entirely. But that’s for another video. Anyway, I think this should be about it. Gaming, gaming and working. And if working, then none of these are good. But I think this should be it for today.
[00:07:43] On a side note, we also have a Discord server, so if you want to join, the link is down below. And of course, we still have channel membership, so if you are looking for a good way to sell yourself an OG poop emoji, that’s a pretty good way to go. Additionally, you can rest assured that the income will not only keep the channel afloat, but it will also serve to fully equip the studio with an army of smoke detectors. Because besides normalizing the degree at which bananas can be bent, the EU now demands at least one smoke detector per core of a Zen 4 CPU during benchmark sessions. And the 7950X has a lot of cores.
[00:08:15] Anyway, thank you for watching and if you want to continue, have a look at the series where we built the new radiator fan testing machine. It was a fun ride. Hope to see you in the next one. Bye bye.