What makes you think that?Holy crap, this Computex is going to be the most obnoxious dumpster fire ever.
View attachment 98468
Lisa - "We're seeing faster ramp up and more designs than Hawk in the same timeframe"
@adroc_thurston "They wouldn’t be getting so many design wins if the product wasn't good."
That's a very strong line-up. Even managed to get some ASUS ProArt wins (Usually Intel dominated). Lenovo, HP, Dell and Acer will probably follow with huge portfolio of design wins for Strix. Computex will be nice.
Its ridiculous.If this is confirmed and gets a reliable, readable scheme, that's it, I'm happy.
Apart from the obvious "AI" moronic sticker, but hey, I'll take "AI HX 170" over "8050HX" any day. Faster, simpler, clearer, just increment the last number to get a refresh out.
All I asked for.
As usual.Here are 17 laptop variants that should promptly be replaced by Strix Halo variants in 2025.
Yet I bet we'll still be seeing OEMs pairing a 40CU Strix Halo with a RTX4060 8GB because people won't think it plays games unless it has a geforce in it.
All things considered, I'm not really sure of why they'd put Strix Point along with a 4050. It's a 20 SM part, it should be a tiny improvement in power. 4060 and up I can understand, but this one is strange.
That's a very strong line-up. Even managed to get some ASUS ProArt wins (Usually Intel dominated). Lenovo, HP, Dell and Acer will probably follow with huge portfolio of design wins for Strix. Computex will be nice.
It is sane and clean, provided that it is indeed Gen - Tier - Rev. The previous one was absurd.Its ridiculous.
It is sane and clean, provided that it is indeed Gen - Tier - Rev. The previous one was absurd.
I'm pleased.
All things considered, I'm not really sure of why they'd put Strix Point along with a 4050. It's a 20 SM part, it should be a tiny improvement in power. 4060 and up I can understand, but this one is strange.
Nah the 4050 still earns it's spot imo. The performance in games should be higher enough to be worth consideration. For the ProArt devices, access to CUDA is probably enough of a selling point on it's own TBH.All things considered, I'm not really sure of why they'd put Strix Point along with a 4050. It's a 20 SM part, it should be a tiny improvement in power. 4060 and up I can understand, but this one is strange.
Not going to repeat myself.It gives less information. Condensing a lot of attributes into a relatively simple name is always a trade off and I think the Year - Tier - Arch - Feature Isolation - TDP bracket is a decent compromise that informs end users while keeping OEMs happy by having a digit they can increment year on year.
20SMs, 6Go, vs RDNA 3.5 16CUs?Nah the 4050 still earns it's spot imo. The performance in games should be higher enough to be worth consideration. For the ProArt devices, access to CUDA is probably enough of a selling point on it's own TBH.
Bro, bro.It’s GeForce bro. Just 20 more SMs bro. OEMs just need a green sticker bro.
AMD learned from Nintendo ninjas and silenced everyone this time.Strange, still no solid numbers yet. The floodgates usually have already been opened. But not this time. Zero, zilch, nada. Weird.
It gives less information. Condensing a lot of attributes into a relatively simple name is always a trade off and I think the Year - Tier - Arch - Feature Isolation - TDP bracket is a decent compromise that informs end users while keeping OEMs happy by having a digit they can increment year on year.
I have some random thoughts:
- 5.1 GHz boost? That's on Phoenix level and very slightly lower than Hawk Point. Not sure there will be an higher clocked part.
- HX moniker seems to point out these will be 45-55W parts. Kinda expected because there are more cores, more capability per core and a better GPU on basically the same process.
- NPU lost 5 TOPS compared to the rumors, maybe for limiting consumption. Not a big deal, but Intel stans are already pointing out that LNL will have "100+ TOPS so Strix Point is useless". When performance should be measured on the field with all applications
- GPU clock is still unknown. Depending on the clock rate we could see from +33% to almost double the theoretical performance of the 780M, of course real performance will be quite BW limited in many scenarios
You might even call it a Halo part...If I had to guess I would suspect that 170 is leaving room for a 180 and 190 part.
Yeah, looks like AMD might be somewhat copying Intel's latest format which will give them the freedom to mix arch. generations into the same model range with no indication of which SKU is what gen. Fun.
LNL is a different weight class, so who cares.- NPU lost 5 TOPS compared to the rumors, maybe for limiting consumption. Not a big deal, but Intel stans are already pointing out that LNL will have "100+ TOPS so Strix Point is useless". When performance should be measured on the field with all applications
You might even call it a Halo part...
Wow everyone switched sides quick.😏
IIRC Zen 5 was originally projected to lose clocks versus Zen 4. Same clocks as PHX is actually quite excellent. It's a much bigger core.I have some random thoughts:
- 5.1 GHz boost? That's on Phoenix level and very slightly lower than Hawk Point. Not sure there will be an higher clocked part.
Frankly 55W on a 12 core with 16 CUs is nothing short of miraculous if it's not power starved.- HX moniker seems to point out these will be 45-55W parts. Kinda expected because there are more cores, more capability per core and a better GPU on basically the same process.
Has it?- NPU lost 5 TOPS compared to the rumors, maybe for limiting consumption. Not a big deal, but Intel stans are already pointing out that LNL will have "100+ TOPS so Strix Point is useless". When performance should be measured on the field with all applications
I expect 33% CUs + 20% broad improvement in clocks (at least). So I'd bet around 60% more perf.- GPU clock is still unknown. Depending on the clock rate we could see from +33% to almost double the theoretical performance of the 780M, of course real performance will be quite BW limited in many scenarios
Just advertise it with the F Zero X OST and you're set to go.Z2 Ultra Extreme XTX would make just as much sense at this point.
I hope I truly get what I want. I just need confirmation that it's really Gen/Tier/Rev. Could just as well be Year/Tier/Rev or something absurd. I don't have any hopes for AMD until I see the presentation.People moan about new things then later realise they actually had something good and regret moaning about it.
From a "certain someone", Strix 16 CU does >4150 on TimeSpy. 12 CU does >3000 TS while using 22 - 23W.GPU clock is still unknown. Depending on the clock rate we could see from +33% to almost double the theoretical performance of the 780M, of course real performance will be quite BW limited in many scenarios
Kepler said on twitter than the clockspeed is wrong. Should be higher.5.1 GHz boost? That's on Phoenix level and very slightly lower than Hawk Point. Not sure there will be an higher clocked part.
Interesting. I don't see in the link what the poster on X shows in the screenshot. Either they fixed the leak or something is fishy. . .
What I see:
View attachment 98521
What was posted:
View attachment 98522
Font seems to be different and they've duplicated the AMD Ryzen AI branding for the CPU name and combined TOPs naming. The row order isn't the same either but maybe they cropped the parts they didn't care about. Not sure, but seems fishy.