I don't understand, if it makes zero difference then why do it? If it doesn't make zero difference then what's the difference in cost and how does that affect the comparison?
I don't understand, if it makes zero difference then why do it? If it doesn't make zero difference then what's the difference in cost and how does that affect the comparison?It's all in the video description. Also the pinned comment -
I think it's because of the hardware he has, and the optimal tuned settings he can achieve for each platform. Silicon lottery and all. I appreciate testing this way, because one of the biggest copes is whinging about properly tuned configurations instead of stock. He is testing in a manner that addresses those excuses on each platform.I don't understand, if it makes zero difference then why do it? If it doesn't make zero difference then what's the difference in cost and how does that affect the comparison?
Because we're not presented with a competition, but rather with proof that one CPU is better than the others. As long as the result is clear, the imbalance matters very little.I don't understand, if it makes zero difference then why do it?
How about athlete's competing in deadlift, and Athlete A lifts 600 lbs and B lifts 500 lbs. B says I could lift way more if I had straps, so they do it again with B using straps. Now A lifts 600 lbs and B lifts 550 lbs.Because we're not presented with a competition, but rather with proof that one CPU is better than the others. As long as the result is clear, the imbalance matters very little.
Allow me a simple analogy: one athlete claims to be better than another at lifting weights. So someone gives the two athletes two weights identical in shape and form, but not mass. Athlete A is given the heavier weight and lifts it up in one try. Athlete B is given the lighter weight and cannot lift it even after multiple tries. The test conditions for the two athletes were obviously different and potentially unfair, but this particular result did not need the weights to be equal in order to show the strength difference between the two competitors. It's not ideal, but somehow good enough considering the circumstances.
It eventualy means that the 9950X3D could be quite better than the 9800X3D given the big driver overhead, or not, we ll see soon.
That reminds me of the arguments for keeping your 7800X3D and not upgrading to the 9800X3D. That you can skip this generation. If we apply that argument, which has been widely and loudly made against both: how big is the gap between 7800X3D+4090 and 9800X3D+5090?I realize this isn't a GPU thread but the 5090 looks kinda pointless, especially when it comes to 1%/.1% lows. 9800X3D + 4090 does almost as well.
Dunno, you'd have to go digging through some reviews to find out. But there must be a fair few 9800X3D owners out there with AIB 4090s with higher-than-normal clocks etc. that will be sitting pretty for awhile, at least in terms of raster performance @ 4k resolution.That reminds me of the arguments for keeping your 7800X3D and not upgrading to the 9800X3D. That you can skip this generation. If we apply that argument, which has been widely and loudly made against both: how big is the gap between 7800X3D+4090 and 9800X3D+5090?
As shocking as the news will be for some, it's driver immaturity on Blackwell. Nvidia told Tom's they are looking into it.But why is the 9800X3D falling behind RPL in ray tracing now?
That seems to be caused mostly by the horrible frame dip in Minecraft.But why is the 9800X3D falling behind RPL in ray tracing now?
View attachment 115547
View attachment 115548
As shocking as the news will be for some, it's driver immaturity on Blackwell. Nvidia told Tom's they are looking into it.
A frame dip only on AMD? How can it be??? Is this some kind of... A.... M.... Dip??????????That seems to be caused mostly by the horrible frame dip in Minecraft.
AFAIK they did not, so make sure the thing is still disabled (NVIDIA App Settings > Features > Overlay > Game Filters and Photo Mode)Reminds me, I should see if Nvidia fixed their issue with the new control panel software. I forget what I had to turn off last month to prevent issues. The software auto updating when I opened it was not nice.
So having extra cores does help with driver overhead, at last on high-end NV cards. There was speculation that higher core counts would also help with Intel A580 driver overhead as well, and while those charts aren't direct evidence, it does seem to hint that they would be useful.As shocking as the news will be for some, it's driver immaturity on Blackwell. Nvidia told Tom's they are looking into it.
Nah, why bother? Both are more than happy to have CPUs absorb the overhead and save them the transistors/effort when its working "as intended".So having extra cores does help with driver overhead, at last on high-end NV cards. There was speculation that higher core counts would also help with Intel A580 driver overhead as well, and while those charts aren't direct evidence, it does seem to hint that they would be useful.
Moral of the story: fix your stupid drivers NV and Intel.
In this case, "faster" is relative. Is a 13900k faster than a 9800X3D? Thanks to crappy NV drivers, apparently so . . . but only for ray tracing.Nah, why bother? Both are more than happy to have CPUs absorb the overhead and save them the transistors/effort when its working "as intended".
Buy a faster CPU
It's Tom's, what did you expect?Anyone else think Tom's was a bit... generous giving this thing 4/5/5 stars?