Something else to keep in mind is AMD’s extraordinarily poor software quality control. I wouldn’t have any concerns purchasing an AMD laptop after reading reviews, but I probably wouldn’t build another computer myself using a newly-released AMD CPU. Zen3 was an ongoing nightmare for weeks and it shared a socket and AGESA with prior Zen CPUs. It’s fine now, of course, I’m only talking about building in the first 3 months.
12th gen Intel changed their entire architecture, process, and chipset. They supported DDR5 too. And everything just worked.
The numbers are weird. The clock boost accounts for about 12% of that 15%, which would indicate very little IPC jump?
That’s correct, and the new AM5 socket can use a ton more power to boot. Doesn’t necessarily mean the 7950X will be a 170w base TDP part, but it’s possible.
Like I said above, the integrated GPU across the board is nice, and I like that the I/O chiplet uses a ton less power at idle. Otherwise Zen4 looks like a pretty iterative release.
My plan is to upgrade my 3700X to a 5800X3D. That’s easily the best bang-for-the-buck upgrade, as I can keep my X570 motherboard and DDR4 RAM, although I may upgrade the latter if there is a significant boost to going up from 3200MHz.
Waste of money really, unless you’re an esports gamer at 1080p. I probably wouldn’t upgrade a 3700X to a 5000-series at all, but if I was going that way I’d get a 5700G or 5900X.
Do you actually game at 1080p? If so then sure.
I’d never heard of the 5800X3D before. What the heck is that? Why is it so much faster than the 5950X for Flight Sim?
It’s an improved 5800X that incorporates 3D cache (cache built on top of the die) to boost the amount of unified L3 cache to a whopping 96MB.
A 3700X has 32MB of cache, but due to the nature of the chiplets it is effectively 16MB of cache per chiplet, and there is a lot of memory latency and inefficiency as the two caches need to talk to one another.
It’s the fastest gaming CPU right now, but not the fastest for anything else. Thing is, the CPU doesn’t matter very much in most games outside of 1080p. MSFS is a pretty extreme outlier in pretty much every benchmark.
Here is how 41 games perform with a 5800x3d at different resolutions:
9% increase from 5800x on average.
Much more dramatic from a 3700X, too. The 5800X at least has a unified L3 cache, just a smaller amount than the 5800X3D.
The split memory cache in much of the 3000-series resulted in very high memory latency.
Basically, the CPU doesn’t matter much at 1440p and really doesn’t matter at 4k, except in specific titles, typically strategy games but also MSFS. In other games your money is much better spent elsewhere.
I doubt even a 4790K would bottleneck much with a RTX3080 running something like Far Cry 6 or Assassin’s Creed Valhalla at 4k.
I didn’t watch this one again but it’s just bizarre how HU do all this work, produce the great comparison graphs, then throw away all the info by taking an average.
In half the games it’s 50% faster, and in half it’s 50% slower, so on average it’s about the same :)
Overall it’ll be better than the 5800X in most games, just not in a way that’s noticeable or that justifies the extra expense unless you game at 1080p.
It’ll be slower than the 5800X in pretty much all non-game activities that don’t leverage the extra cache.
Well, if you watch it you will see they also show 15 individual game comparisons and then a graph with the 41 of them, rating the % difference:
Bunch of info emerging about upcoming gen. Mild mannered Australians hold forth here :)
rei
1605
Everyone hopeful about Intel’s Xe remember the constant incompatibilities and bugs their iGPUs have suffered with each new game in the last decade. Usually people trying to run games off laptop Intel iGPUs.
They need to have beefed up their developer support and drivers teams by at least ten fold.
I’m hopeful over the long term that another competitor in the low and mid tiers can help bring prices down. At the very least, more volume of passable cards at competitive prices in the current environment would be something.
