He normalized it to system power because when set at the same SoC power the AMD system was using more power. While there's contentions regarding the methodology, if it was say in a handheld system, it would make sense.The Strix Point SOC in those games is running with a way lower power limit. The GPU is running at less than half the frequency of the Xe GPU in CS2, it's hard to make any real comparison off that with them so far apart.
True, but Alchemist is N6, so still a full node advancement. And the current rumors are while the A770 is 225W, the peak B version might go to 300W. With clocks said to be solidly in the 3GHz range, the performance potential is there.Different process, different power limits. How competitive it is depends on the v/f curves, bandwidth, pricing and so on relative to the competition.
He normalized it to system power because when set at the same SoC power the AMD system was using more power. While there's contentions regarding the methodology, if it was say in a handheld system, it would make sense.
wow, thats disappointing. wonder why Apple kept the same media engines for 4 ***** years if its not class leadingI'm surprised how relatively bad Apple is at this.
I don't get this why can't Intel launch all 2 dies at the same time? Unlike others Intel needs to go all out not stagger releasesIf the timeframe for Battlemage is really B580 in December with 4060 performance and G31 die in middle of next year, it greatly increases the chances of them cancelling Celestial dGPU.
I don't get this why can't Intel launch all 2 dies at the same time? Unlike others Intel needs to go all out not stagger releases
Because, silicon is hard, and if you compare to a building, Intel has been building 2-3 story ones, while AMD/Nvidia has been making skyscrapers. Now Intel has to figure out how to get there. It is not copy and paste making bigger chips, just as skyscrapers aren't 3 story buildings multiplied and stacked on top of each other.I don't get this why can't Intel launch all 2 dies at the same time? Unlike others Intel needs to go all out not stagger releases
Because, silicon is hard, and if you compare to a building, Intel has been building 2-3 story ones,
I agree but I'm worried if they delay past december then it will compete with RDNA4Maybe G31 needed extra time to get those 3GHz+ clocks, or maybe it's yield at that frequency, or a combo of both. The bigger die must be justified with proportional performance improvement.
That's the thing. Battlemage is so close to release it's showing it on retail sites, yet they haven't even talked about it. So they feel bad about something. The only other alternative is that it's really good and kept it hushed like Core 2, but what are the chances of that for Battlemage?I agree but I'm worried if they delay past december then it will compete with RDNA4
The OpenCL score of 78,743 points is clearly not high enough to be considered final. In fact, it is lower than the Arc A580, which is expected to be a slower card.
| Device | Score |
|---|---|
| Intel(R) Arc(TM) A580 Graphics | 89699 |
| NVIDIA GeForce RTX 2080 Super with Max-Q Design | 89538 |
| NVIDIA GeForce RTX 3060 | 88994 |
| NVIDIA GeForce RTX 4060 Laptop GPU | 88915 |
B580 OpenCL final Result should be around 4060 TiBMG GB Opencl benchmark leak
B580 initial result is worse than A580
Device Score Intel(R) Arc(TM) A580 Graphics 89699 NVIDIA GeForce RTX 2080 Super with Max-Q Design 89538 NVIDIA GeForce RTX 3060 88994 NVIDIA GeForce RTX 4060 Laptop GPU 88915
One thing the graph is not showing is power usage. It may well be the case that for Apple it was a trade off and they erred on the side of efficiency.wow, thats disappointing. wonder why Apple kept the same media engines for 4 ***** years if its not class leading
Does it? Because OpenCL performance isn't same as gaming.B580 OpenCL final Result should be around 4060 Ti
I meant OpenCL. It should score 130K which is 4060 Ti level.Does it? Because OpenCL performance isn't same as gaming.
Maybe with 24Xe2 variantI meant OpenCL. It should score 130K which is 4060 Ti level.
So targeting 4060 performance most likely.
Honestly, LFG. A solid 12GB card at that less than $300 price point, it’ll be up against the 3060 12GB and I hope it clobbers it.So targeting 4060 performance most likely.
Especially in RT.A solid 12GB card at that less than $300 price point, it’ll be up against the 3060 12GB and I hope it clobbers it.
LFG ? Looking for group anandtech is a game lobby 🤔Honestly, LFG. A solid 12GB card at that less than $300 price point, it’ll be up against the 3060 12GB and I hope it clobbers it.
Umm, I'm using a more contemporary definition by Google's explanation.LFG ? Looking for group anandtech is a game lobby 🤔