- Mar 3, 2017
- 1,747
- 6,598
- 136
Nahh.. I think the 65W 9000s will be the budget kings since the A620 boards supports 65W cpu's.The "budget" AM5 option is now the 7000 series, the new thing is 9000 and the hot stuff will the the 9000 X3D chips.
End of July. I think both GNR and STX NDAs lift then, if I'm remembering correctly.
It is correct.Just one data point, that may or may not be correct.
That's because Strix ships before GNR.I have seen a Best Buy page for a Strix Point laptop that said release date of July 15.
Yeah, ~22% faster than PHX in 1t in cinememe, that's more in line with let's just say our findings.
1T on par with 7950X. nT on par with 5950X.
The GPU is very powerful.
Can we get a real int benchmark not cinebench.....plz god!!!!!
1T on par with 7950X. nT on par with 5950X.
The GPU is very powerful.
You already have a 3.7 GB6 run, look at the INT subtests (HTML5 etc) and scale with freq.Can we get a real int benchmark not cinebench.....plz god!!!!!
51% PPW, we are so back Radeon bros.
1T on par with 7950X. nT on par with 5950X.
The GPU is very powerful.
Given the fact Zen4 is more or less equal in performance to Raptor Lake and Zen5 being supposedly only 15 percent above it on average, i would not call this "sticking it to Intel".This is just all about sticking it to Intel, as, really they can; these chips are not expensive to produce and you get a lot of PC even with a B650 board.
Yes it actually is.Is it really?
Didn't know that, thanks!GPUs are a complex piece of silicon that needs compute units, texture units, graphics memory for the highest detail overview and many countless other small details.
And especially if BW is still limited by the RAM. It must be seen, if that is the average improvement, or a single data point. What it is interesting is that at the same power, the 890M has +51% improvement in Timespy score respect to the 780M.So increasing CU and getting 36% is quite impressive, especially if the change in process technology is not big.
so lets compare at "sane" levels for mobile CPUIs it really?
Another thing is that GPU's primary way of increasing performance is more parallelism, which you cannot do for CPUs, because graphics are massively parallel. So the main focus is increasing more shaders and make sure they scale well.Didn't know that, thanks!
The thing is, TS graphics score correlates perfectly with dGPU gaming performance, which is not the case with APUs, if the claimed gain of 36% is true.And especially if BW is still limited by the RAM. It must be seen, if that is the average improvement, or a single data point. What it is interesting is that at the same power, the 890M has +51% improvement in Timespy score respect to the 780M.
780M's shared memory run at what, 5200mt/s?Combined with LPDDR5x 7500 MT/s memory, the memory bandwidth is simultaneously increased
Did I miss something because afaik there isn't oneYou already have a 3.7 GB6 run, look at the INT subtests (HTML5 etc) and scale with freq.
GPD Duo indicated on the test as 780M reference uses the same, LPDDR5x 7500780M's shared memory run at what, 5200mt/s?
These are excellent levels of performance for the iGPU. You're basically matching an RTX 3050/GTX 1660 but in an integrated format factor.
At 45W?
1T on par with 7950X. nT on par with 5950X.
The GPU is very powerful.
Did I miss something because afaik there isn't one
It looks like RDNA3.5 is a bigger increase over RDNA3 then RDNA3 was over RDNA2.so lets compare at "sane" levels for mobile CPU
54w Time Spy:
2791 / 4221 = 1.51
Phoenix shows diminishing returns if it is allowed to go up to 72W almost as it it was a desktop APU:
72 / 54 = 1.33 power increase, 3218 / 2791 = 1.152 score increase
Imaginary Phoenix with glued on 4CUs that magically do not consume any energy (still at 54W):
2791 * 1.33 = 3712
Still can't beat or even match it...
3712 / 4221 = 1.13 >> something in RDNA 3.5 consumes significantly less power (might not be the CUs...)