The replacement cycle being lengthened for desktops is not the same as "practically nobody buys a desktop anymore".
I know I'll eventually replace/update my desktop, won't you do the same?
I will, but I'm not the typical consumer.
^ Desktop Skylake i3's and above are fairly close to AMD's iGPUs, on average, from what I've seen. There are still cases where AMD pulls ahead significantly, most likely because Intel doesn't spend nearly as much time optimizing for specific games. If Skylake's iGPU is "mediocre", then so is AMD's... and I'd probably argue that, yes, they're both mediocre.
If comparing a Celeron/Pentium without a discrete GPU vs an A6 or A8, then yes, I'd say the AMD CPUs will have a broader range of usefulness. There are plenty of games where AMD APUs simply cannot provide a playable framerate at all, regardless of settings, but the number of games will be smaller than those chips equipped with, say, Intel's 510 iGPU.
Which setup would be better for SF V? An i3-6100 with HD530, or a G1820 with a 2GB GTX950?
Well 1211 to 1908 is a ~33% difference and since any game, no matter how well threaded it is, will have at least one thread that will benefit heavily from running faster this does make a big difference.I know this benchmark isnt everything (or anything) but I'm not seeing some mind boggling difference, especially when you consider the 8320E with a decent cooler cost me $150 (55c stress tested as shown)
Well 1211 to 1908 is a ~33% difference and since any game, no matter how well threaded it is, will have at least one thread that will benefit heavily from running faster this does make a big difference.
Intel already fixed this bug and many MB vendors already included the fix in their BIOSes. You can stop trolling with this news now.
They don't have an Intel Shrine: they have the AMD chamber of Torture instead.Well you can stop trolling for Intel while you're at it -- but I seriously doubt that will happen either.
Seriously, I run a Devil's Canyon -- but I don't have an Intel shrine in my bedroom like many on this forum.
For what it's worth, I don't think I own a game I play today that is not GPU-bound a big majority of the time. Even Grim Dawn (small developer, uses a single core from what I see) is GPU bound at my middling 19x12 resolution with my mildly overclocked 7970.
That 33% IPC disadvantage gets to go against i3's that cant be OC'd
Then don't compare yourself to an i5,if you'r gonna be GPU limited anyway than it doesn't matter if you have an fx-4xxx a i3 or even a pentium.
If your real, practical use is with your CPU at 100% all the time then sure,but how real and practical is that for the average joe?Actually, I will compare it to the i5. In a thread about real, practical use,
Yes,yes in deed, you do have to have a fast enough CPU...You have to have a fast enough CPU to be GPU bound, otherwise you'd be CPU-limited...
lmfao, in what kind of bizarro world does having an IPC advantage equate to being a disadvantage?
Can you relate clock speed to performance?
lmfao, in what kind of bizarro world does having an IPC advantage equate to being a disadvantage?
lmfao, in what kind of bizarro world does having an IPC advantage equate to being a disadvantage?
The bug is fixed, any Skylake user testing with Prime95 and other similar loads knows.
And no performance penalty either. Thought luck the ADF cant use it anymore. It was fixed very fast, perhaps that why some people are so behind the curve
Intel already fixed this bug and many MB vendors already included the fix in their BIOSes. You can stop trolling with this news now.
I guess removing overclocking doesn't count as a performance penalty by your standards. LOL
It's a huge disadvantage if you've only got 2 threads for heavy multithreaded apps/games (versus a slower IPC quad). All the IPC in the world couldn't make up for my G3258 which was a dual core living in the now-quad core world -- way too many games stuttered because a mere 2 cores just can't run multithreaded games well.
No, the locking is 2 steps above, 0x76 is got the lock, 0x74 fixes the bug.
It was removed in a later microcode version 0x76, and only for non K CPUs. Something that was never officially allowed. The bug was fixed in 0x74. But I can understand your confusion when you dont follow it so you end up mixing it together to make a wrong conclusion.