- Mar 3, 2017
- 1,747
- 6,598
- 136
No 9950x ??Theres new multiple new runs of 9900X, 9700X, and 9600X on the Geekbench browser for today. All seem to be from MSI systems.
Based on all of these being done on MSI boards and the time stamp of when the tests were run it looks like Hardware Unboxed received their Zen 5 review samples.Theres new multiple new runs of 9900X, 9700X, and 9600X on the Geekbench browser for today. All seem to be from MSI systems.
For comparison sake 13900K can do 30500 at 100W PL2.
Manually undervolted or just set PL and auto voltage?For comparison sake 13900K can do 30500 at 100W PL2.
the Kavinsky water leaks 💦"Kavinsky" milking that single ES dry, forum dwellers starting to taste less milk and mostly water. But it is alright, folks have done worse things to attain the "leaker" badge
I have seen claims that R24 favours x64 by supporting AVX2/512 but Chips&Cheese review states that it is only using AVX encoding for mostly scalar operations. Only 7% of instructions are using 128b. Therefore most of the register width is unusued [32b arguments, vs 512b register width in extreme case]. According to the review what has changed is that the working set size is too big to fit in LLC so it stresses memory subsystem more compared to R23. In fact it seems that limiting it to mostly scalar operations favour cores that have large number of smaller fp units, as the size of the unit will not make a difference. So whatever gain Zen5 will have over Zen4 will come from improvements to the front-end, load store unit and integer side of the benchmark as the fp side will most likely not see any gains.R24 also is also better for ARM too. It makes use of the cores better than R23 did.
It seems to me AMD did quite well, achieving increased density while using a minor node transition and keeping max clocks intact. I'm very curious to see Zen 5C die area.Some info from Zen5 tech day: [...] CCD Transistor density +26.8%
Definitely. Just don't ask them about RDNA3.So AMD has very good Physical Design capabilities?
I wonder if "lot more different" should be taken in positive or negative way... could DDR bandwidth be insufficient or would power draw on 16 core vs 12 core be a problemSomebody else has a 9950x not an es sample aswell
I wonder if "lot more different" should be taken in positive or negative way... could DDR bandwidth be insufficient or would power draw on 16 core vs 12 core be a problem
Zen 5 mobile seem to have worse performance per clock vs desktop zen 5 in geekbench 6. Hx 370: 2860@5.1ghzThrowing this on the radar. Would be nice if there was a boost clock jump if this HX 375 is for real. Only an NPU jump to 55 TOPs as of now.
HP's upcoming laptop to feature Ryzen AI 300 processor with 55 TOPS NPU, surpassing HX 370 SKU in speed - VideoCardz.com
HP reveals OmniBook Ultra with the fastest NPU on the market, another variant of AMD Ryzen AI 300 incoming? HP has issued a press release claiming to have the fastest Neural Processing Unit (NPU) on the market. They describe their upcoming laptop as “the world’s highest performance AI PC with up...videocardz.com
Throwing this on the radar. Would be nice if there was a boost clock jump if this HX 375 is for real. Only an NPU jump to 55 TOPs as of now.
HP's upcoming laptop to feature Ryzen AI 300 processor with 55 TOPS NPU, surpassing HX 370 SKU in speed - VideoCardz.com
HP reveals OmniBook Ultra with the fastest NPU on the market, another variant of AMD Ryzen AI 300 incoming? HP has issued a press release claiming to have the fastest Neural Processing Unit (NPU) on the market. They describe their upcoming laptop as “the world’s highest performance AI PC with up...videocardz.com
Clocks still fluctuate for whatever reason, it's @ 4.8 most of the time. Not sure if it's a reporting issue or the thing can't hit boost clocks consistently.Zen 5 mobile seem to have worse performance per clock vs desktop zen 5 in geekbench 6. Hx 370: 2860@5.1ghz
No way zen5c is hitting 4.8 GHz so the clock reporting is definitely off...Clocks still fluctuate for whatever reason, it's @ 4.8 most of the time. Not sure if it's a reporting issue or the thing can't hit boost clocks consistently.
"frequencies": [
4837,
4839,
4839,
4837,
4838,
4802,
4834,
4840,
4716,
4837,
4839,
4837,
4840,
4838,
4836,
4556,
4840,
4839,
4836,
4839,
4839,
4840,
4836,
4838,
4836,
4547,
4771,
4841,
4839,
4837,
4839,
4840,
4841,
4838,
4839,
4687,
4727,
4852,
4910,
4978,
5045,
5116,
5142,
5141,
5140,
5139,
5141,
4988,
5140
]
GNR does not exhibit this kind of behaviour.
Could be surely, in context it would seem that it should be a 370.I think that's just a typo. There are a lot of errors in these placeholder websites. ASUS also has weird errors like that on their strix point laptop pages.
That’s the key question because a 7945HX does 33400.Someone got 22.2K MT with a Strix 370, per @HXL@9550pro
No power consumption figures shown.
pic.twitter.com/BXsKxYfd1a
View attachment 103229
A 7945HX is for a different market compared to a Ryzen 370AI. It has (can have) different power envelope, it is a desktop derivative with 16 cores explicitely dedicated to gaming desktop replacement laptops. Its succesor will be the Fire Range lineup.That’s the key question because a 7945HX does 33400.