Quote from: M4M on September 19, 2025, 23:49:45

Quote from: Worgarthe on August 30, 2025, 23:14:59

Quote from: M4M on August 30, 2025, 23:08:25M4
Five NEON FMLA ops per clock cycle. Do the math Re@tard.
And st@fu you aint talking sh@it IRL
weak inc@el Fu@ck

Aaah, ok, ok, I get it now... So you write lkfhusdgfvncsiuvfqrh oisf hnouiqeswh nfjhr because of being in a hurry as your mom is trying, for your own good, to prevent you from using a phone or other tech. So before she comes back you just smash random stuff on a keyboard.

Thinktrash imbecile take the fucking shower until than stop fucking yapping to me incel rat! Just come up to me IRL soy i will fucking crush your jaw the second you enter my aura!

You are very slow to reply. Meaning your Apple chip is too slow. Rip.

Quote from: MM128 on August 30, 2025, 12:48:13@M4M
I tried your "C-RAY is the non biased most accurate version of CB R23".

So, while the CB R23 gives the same result after "warming up" (CPU throttling) the system with a difference of 1-2% each and every time, this pile of s..... shows results with a difference of 20 even up to 25%!!!
I'm 100% sure that some iMoron like you programmed it, lol

because you re a biological muller x86 shill and you dont even know basic compile flags.
Go f*** yourself x86imbecile or else i will for everyone to see what a soy bitch you are

Quote from: M4M on August 30, 2025, 23:08:25M4
Five NEON FMLA ops per clock cycle. Do the math Re@tard.
And st@fu you aint talking sh@it IRL
weak inc@el Fu@ck

Aaah, ok, ok, I get it now... So you write lkfhusdgfvncsiuvfqrh oisf hnouiqeswh nfjhr because of being in a hurry as your mom is trying, for your own good, to prevent you from using a phone or other tech. So before she comes back you just smash random stuff on a keyboard.

Quote from: M4M on August 30, 2025, 19:53:16And dont even start arguing about it little IN"C"EL
You aint doing s*** when you see me, you be running like a pray i promise you that! TAKE THE f***"ING SHOWER I CANT STAND YOUR DISGUSTING FI"LFTH a**, DONT F"UC"CKING TALK TO ME UNTIL YOU TAKE THE SHOWER AND GET SOME TOOTHPASTE TOO!! IN"CEL F"U"CK!!!!!

LOL, you're funny.

Brainrot is real with this one.

@M4M
I tried your "C-RAY is the non biased most accurate version of CB R23".

So, while the CB R23 gives the same result after "warming up" (CPU throttling) the system with a difference of 1-2% each and every time, this pile of s..... shows results with a difference of 20 even up to 25%!!!
I'm 100% sure that some iMoron like you programmed it, lol

@M4M
"...we alpha athletic charming leaders, managers, doctors, politicians dont hear your tiny pathetic Incel voice. You dont matter!"

Well, your post is an "alpha pr.ck" post, that is for sure and with and with a 10% market share it's pretty hard to say that you matter :D:D:D

Quote from: M4M on August 25, 2025, 12:08:28M4PRO and M4MAX smokes this and its proven by the goldstandard FP Throughput Benchmark
M4MAX is beyond any mobile chip and destroys the most powerful consumer DESKTOP CPU

Maybe, but the Apple-World is a very restrictive ecosystem, which is reason enough to avoid by most people.

Quote from: M4M on August 25, 2025, 12:08:28M4PRO and M4MAX smokes this

The M4 machines can't do audio-pass through, so no Atmos, no DTS:X, no go for me.

Quote from: M4M on August 26, 2025, 12:06:23This is a SoC for professionals, not clowns who game all day, are fat, weak, and ugly idiots who have the personality of a 5year old toddler.
Incels opinion dont matter, we alpha athletic charming leaders, managers, doctors, politicians dont hear your tiny pathetic Incel voice. You dont matter!

I beg to differ, I'm going to use my Framework Desktop as a silent but powerful home-cinema + gaming machine. I might play with AI a bit for fun.

You seem thoroughly invested in your mental image of who will be buying these PCs. It doesn't seem healthy.

I'd recommend that you get some fresh air, touch grass, meet some friends...

Quote from: Worgarthe on August 19, 2025, 00:38:24

Quote from: heffeque on August 19, 2025, 00:29:17

Quote from: beerkegbob on August 18, 2025, 16:17:21For $1400 or $1600 I could have a full size gaming desktop.
Why would i pay that for something that limits my options?

Why would I want a huge desktop PC when I can have a compact and efficient PC?

Repairability? Upgradability? Longer warranty on each component inside? Non-soldered RAM but instead upgradable up to 192 GB? Full control of any item inside? 9950X is faster than this CPU? It is also overclockable to squeeze even more performance? RTX 5070 Ti stomps this 8060S (exactly twice faster in benchmarks, much faster in games and it can run almost any game in ultra/maxed at 1440p at 165 Hz native without VRR, no need for upscaling either)? Quieter with better cooling (larger case, larger vents)? Fractal Terra is small and stylish case where everything mentioned can fit inside and aesthetically looks much nicer than this here?

All of that is exactly US$1618.02.

This one has LLM workload as its only advantage. Well unless you want to have as small as possible but no idea why that matters to such degree where 10 cm here and there plays any difference, unless you live in a 6 m2 apartment (six, yes).

This is a SoC for professionals, not clowns who game all day, are fat, weak, and ugly idiots who have the personality of a 5year old toddler.
Incels opinion dont matter, we alpha athletic charming leaders, managers, doctors, politicians dont hear your tiny pathetic Incel voice. You dont matter!

M4PRO and M4MAX smokes this and its proven by the goldstandard FP Throughput Benchmark, C-RAY.
M4MAX is beyond any mobile chip and destroys the most powerful consumer DESKTOP CPU, 285K.
M4MAX completes C-RAY 2.0 5K in 116seconds, while M4PRO needs 143sec,
Fastest 285K needs 137sec. 9950X3D is nowhere near at 163sec best run.
C-RAY is the non biased most accurate version of CB R23, as it measures pure FP Throughput and before any r e t a r d shill says its memory bound, no its not.
Ray Casting(here 16rays per pixel) is pure SIMD FP + some integer math. Maxxing out the TDP, stressing the CPU more than R23 as the Power Consumption is Higher than in R23.

For M4PRO its 45w in R23, and 63.5W(presumably hardwired PL1 limit) in C-RAY 5K Image render.
M4MAX consumes about 81w on laptop due to throttle, in mac studio its higher.
Than again in R23 it is only about 50w for M4MAX.

Lets not talk about M4MAX GPU perf, as that would be a humiliation. M4MAX GPU surpasses 150w 4090M in DGEMM(Raw Performance predictor(just like C-RAY for CPU's)) and in other optimized benchmarks like GFXbench.

All in all for the price this AI MAX+ is a good trade but dont ever think it's faster than M4MAX or even M4PRO just because you all jerk off to a biased anti ARM benchmark called R23.
C-RAY has no SSE or AVX inline loops or Intel Embree unlike R23 which is full of garbage X86 intrins and base lib is Embree. CLOWN BENCHMARK.

C-RAY is compiled before able to run it and thus it runs native and utilizes every ability of the maschine on both sides. x86 and ARM

Not really in my ballpark of what I need but I feel you do have to accept some compromises with these small form PC's. Some of with is upgradable parts and potentially some longevity issues depending on cooling effectiveness. Certainly understand not everyone wants a giant PC case sitting on desk or floor. But as with laptops your ability to maintain and upgrade is limited. If I am going to spend some serious money on what I hope is a long life PC. I do prefer to have upgrade option to refresh it and keep it being useful longer. Not to mention all the parts can be cooled more effectively. Which in my view would also contribute to a longer lifespan.

Quote from: Worgarthe on August 21, 2025, 13:39:52Never said that, but you are getting very emotional for some reason. And lol, grab a Framework then and enjoy its ultimate hyper-supremacy over anything else on the market. There is no better PC in the existence.

Just respoding to your emotions with the same level of emotions. I'll tune it down, no worries.
I did grab a Framework Desktop (months ago), it'll arrive on Monday.
And yes, currently there is no better PC in existence for my specific use case, so I'll enjoy it very much, thank you :-)

Quote from: heffeque on August 21, 2025, 00:58:02Please tell me more about how yours are the right ones, and everyone else's are wrong!

Never said that, but you are getting very emotional for some reason. And lol, grab a Framework then and enjoy its ultimate hyper-supremacy over anything else on the market. There is no better PC in the existence.

"Strix Halo" is a 256-bit chip with a theoretical memory bandwidth of 256 GB/s, comparable to a quad-channel desktop PC, except there are no quad-channel desktop PCs, only workstations. Normal/most desktop PCs are dual-channel, hence the higher price and in this small size (it even rhymes, so it must be true). If you don't need the small size and quad-channel for LLMs (AMD specifically advertises "Strix Halo" for LLM inferencing), obviously get a desktop PC the (a desktop PC is not going to be able to replace Strix Halo's 128GB and 256GB/s RAM and AMD advertises this).

I thought I read that on the other language version the cooling didn't perform as well as on this english version of the article? Well, maybe it was a different unit.