AMD ROCm is now on Windows! RDNA2 & RDNA3 GPUs Included!

Get 25% discount on "Gvgmall Back to School" with my "SKAG" code!

Sponsored by Mediamz

Hello Dear Viewers :D

Today, bringing you a video about the official release of AMD's ROCm for WINDOWS 10 & 11 22H2! Supporting RX 6000 & 7000 Series.

00:00 - Why did NVIDIA maintain their dominance of the GPU market?
01:10 - AMD ROCm 5.5.1 (Windows Support)
04:40 - Conclusion

05:49 - Channel Members
06:10 - More Videos

Hope you enjoy the video. Comment with your thoughts about it. Don't forget to hit like, subscribe and share the video if you can.

If you want, support me on:

Social Networks:

AMD AM4 PC USED for CPU testing:

CPU Cooler: Mars Gaming ML360
MB: Asus X570 Strix-F
RAM: 2 x 8GB - Patriot Viper 4400Mhz CL19 (3600MHz CL16)
SSD (Nvme): 500GB + 1TB
PSU: BitFenix Whisper 650W+ Gold
CASE: Bitfenix Nova Mesh SE TG

INTEL 10/11th Gen PC USED for CPU testing:

CPU Cooler: Artic Freezer Penta 33
MB: MSI Z490 Tomahawk
RAM: 2 x 8GB - Patriot Viper 4400Mhz CL19 (3200 MHz CL14)
SSD: Kingston V300 120Gb
HDD: Seagate Barracuda 2Tb 7200 rpm
PSU: Seasonic SIII 650W Bronze
CASE: Aerocool One Frost Edition

INTEL 12th Gen PC USED for CPU testing:

CPU Cooler: Mars Gaming ML360
MB: Gigabyte Z690 Aorus ELITE (DDR4)
RAM: 2 x 8GB - Patriot Viper 4400Mhz CL19 (3600 MHz CL16)
SSD: 240GB
SSD (Nvme): 500GB
HDD: Seagate Barracuda 2TB 7200 rpm
PSU: Aerocool X-Strike 800W 80+ Silver Semi Modular
CASE: Phanteks P400A White

AMD AM5 PC USED for CPU testing:

CPU Cooler: Enermax Aquafusion 360
MB: Asrock B650E Taichi
RAM: 2 x 16GB - Flare X5 6000MHz CL32 (6000 MHz CL30)
SSD (Nvme): 1TB
PSU: Enermax Revolution D.F.X 1050W (80+ Gold)
CASE: Enermax StarryKnight SK30

My Main PC (used for GPU testing):

CPU: AMD Ryzen 7 7700X (5.4GHz Static Overclock)
CPU Cooler: Noctua NH-D15S
MB: MSI X670E Carbon Wifi
RAM: 2 x 16GB - Flare X5 6000MHz CL32 (6000 MHz CL30)
SSD (Nvme): 500GB + 2TB + 2TB + 4TB
PSU: Corsair RM750X (White) 750W 80+ Gold
CASE: Thermaltake View 51 TG

#rx7900xtx #CUDA #amddrivers

[NipahAllDay]

A bit of an update, but AMD's performance on Blender has gotten better in 3.6, since it has HIP RT support. Think of Optix, but for AMD. It's not the best, but it's much better than before.

[cosmicusstardust]

I really hope this is enough for AMD to finally break up Nvidia's monopoly in this market segment

[AncientGameplays]

ROCm is finally supported on WINDOWS for RDNA2 (RX 6000) and RDNA3 (RX 7000) GPUs! Meaning software performance will get better quite soon, most likely starting with Blender.

[Atilolzz]

Fine Wine, guys
Feels good to be a 6800XT owner

[mister_]

I had bought a PC with an AMD Radeon and later started Python and ML courses. What a huge regret with AMD! it is pathetic that I still cannot properly run tensorflow and pytorch on my Windows 11 PC. ROMc for only 7000x series now?? in 2023? are you kidding? I am perplexed to see you are so artificially excited.

[Agp]

they really need to do an explanation on why they take SOOO many years to get support for ROCm to Windows 10 and 11 🤷‍♂

[yoiqsrambadian]

This is a game changer, especially if AMD also support their mobile APU. That'll be great for light mobile workstation.

[evergaolbird]

What I notice about Adobe is how their 2023 builds of their apps has been far better in RDNA2 (I have RX 6700 XT), so I cannot imagine the performance benefits ROCm support can do now in a couple of further updates can do. Looking forward seeing it action real soon.

Adobe products such as Premiere, Photoshop, especially Substance 3D Painter are natively benefiting in NVIDIA CUDA support. So having them natively support ROCm will be a game changer. This will hurt further of me transitioning to Linux as my daily driver too.

I've personally seen the performance difference between two natively available Software Apps in both Windows and Linux with the latter having native support for AMD ROCm (Blender in Linux) and the performance difference between the two platforms of the same app on an AMD GPU was enough to show how good was it from actual workflow/rendering and to even just booting up the app with 10-15 seconds load time difference favoring Linux. Which what some of us got really enticed to completely move away from Windows.

Hopefully ROCm could make a change over that (at least with workflow performance since Linux optimization versus Windows is another big difference due to the extensive community support).

[BlueberryJamPie]

I hope to God it takes off and all the programs that use Cuda also add ROCm support. Actual competition in rendering market is a godsend. Kinda sucks that it's 6800 and above for the whole thing, but at least it's a step.

[altaccount]

Ain't no way I can finally start working with AI on my PC, let's

[ShadowCatGambit]

Another reason why AMD is becoming a better option.

[stevenwest]

Oh I'm glad somebody covered this news, as I was seriously considering buying a 3060 12gb, but now I'll step up to a more logical RX 6800, for a balance in gaming at 1440p and stable diffusion. The Rtx 3080 12gb just isn't cheap enough for the longevity of it.

[victormorales]

I think the word you were looking for was "transcribe"...

[pasananthony]

I haven't had an Nvidia GPU for reference, but I've been using blender on my 6900XT with no complaints. Cycles runs okay on HiP as it is, so if it's gonna get even better I'm hyped!

[Alejandro.budget.gaming]

I will wait a Windows stable diffusion port with ROCm support

[coni]

I don't know why media stayed silent, but since i've updated to Blender 3.6, I noticed 2-4 times performance uplift using HIP on my RX 6700 XT.

[palomito]

I was thinking on buying a RTX just for blender but now I'm keeping on AMD side

[hassanaoude]

Finally! Thanks for the update, time to put my language models to work.

[EpisodicGamer]

FINALLY! AMD GPUs can thrive in AI such as Stable Diffusion and AMD owners can finally generate images as fast as Nvidia in that field.

[duladrop]

I think the accurate word for that is HIP CUDA remapping because AMD uses wave matrix and if they want to use application that only supported CUDA they need HIP to remap wave matrix(ROCm) to warp matrix(CUDA) but this in a sense create remapping penalty performance. That is why I am waiting for application to directly support ROCm instead of just CUDA, so that AMD GPU can really perform without any remapping performance penalties from HIP-CUDA.