I don't think AMD needs to support 5+ year old GPUs personally. And all the recent generations are already practically supported. AMD only claims support for a select few GPUs, but in my testing I find all the GPUs work…
You can paper launch as much as you want but Blackwell isn't shipping until Q4. Also Blackwell's lead will be short lived, because mi350x is coming out next year, and it will have a node and architecture advantage. So…
None of those things matter, if all you're looking to do is run your existing workloads on mi300x in the cloud. You get more bang per buck by going AMD.
ROCm has improved a lot. And you can rent mi300x in the cloud now. So if you have a program that runs on Nvidia GPUs, it takes no time to test it on a cloud mi300x. If it works you can use it and save some money in the…
I don't understand why they should use TensorRT. vLLM is much more popular and it was actually written for Nvidia. It also supports AMD hardware, so it's the appropriate tool to compare.
AMD doesn't have a higher PE ratio. You should use nonGAAP numbers because the GAAP numbers include Xilinx Googwill Amortization which skew the PE. AMD's PE is ~55. Nvidia's PE is above 70.
People used the same argument when saying AMD would never beat Intel in CPUs. Intel has a lot of software engineers. Also these days AMD has a good number of software folks, thanks to the Xilinx acquisition and the…
For us hobbyists used 3090 or new 7900xtx seem to be the way. But even then you still need to build a machine with 3 or 4 of these GPUs to get enough VRAM to play with big models.
zendnn is for CPUs. Pytorch ROCm version doesn't require you to change a single line of code. So all the same semantics like `device = "cuda"` work the same when using AMD GPUs.
libraries if they are popular in general are expected to work with a wide variety of code. Supporting code smell is a feature.
The author also says that AMD doesn't have Tensor cores. No shit, Tensor core is an Nvidia trademark. AMD has matrix multiplication units in their CDNA GPUs.
As a hobbyist you should look into spot instances in cloud. Way cheaper and faster than having to invest in your own hardware.
AMD has been improving a lot. Also you're comparing 3090ti to a 2 generation older RVII how is that a fair comparison?
Yes you can. And AMD's Linux driver is much better.
AMD's driver is much better on Linux than Nvidia's. And ROCm HIP stuff isn't obscure. It's what's needed to bridge Nvidia's proprietary vendor lock ins. Nvidia is literally cancer in this space.
There is also an article about how they did submit a score from one of these Chinese "exaflop" systems, for a different benchmark and it turns out it can only achieve the claimed performance at half precision:…
Epyc 2 (Rome) will have up to 64 cores.
> I guess AM4 also means no real improvements on the PCIe lane count The new Ryzen 3000 CPUs support PCIe Gen4.. so while the number of lanes will remain the same, their bandwidth could be doubled. Just announced Navi…
But the GC has to do it in a thread safe way which involves locking/synchronization. Otherwise you get nasty race conditions.
It's only a space heater when it's being used. These CPUs consume very little power when idle. So if you have a task that can leverage 32 cores.. that task should in theory complete that much faster than on your 65W…
Particularly in case of Threadripper, they use cherry picked binned cores so they can actually hit even [slightly] higher clocks than their mainstream cousins. We don't know the Threadripper 2's clocks yet, but last gen…
It's not happening because it's hard to do. Extracting instruction level parallelism is difficult(which is how CPUs gain IPC) and then also the cores have to be wider with exponential complexity. ie. At the point you…
Let's not forget Raven Ridge (Ryzen Mobile).. It targets lower power and lower performance, but it's still 3x faster than Intel's current solutions when it comes to graphics.
It's important to note their ARM CPUs aren't anything to write home about. Mostly a slightly modified version of ARM's own reference designs.
Intel has paid Nvidia billions in the past to use Nvidia's tech at the expense of AMD.. https://newsroom.intel.com/news-releases/intel-announces-new... Since both Nvidia and Intel are monopolies in their respective…
I don't think AMD needs to support 5+ year old GPUs personally. And all the recent generations are already practically supported. AMD only claims support for a select few GPUs, but in my testing I find all the GPUs work…
You can paper launch as much as you want but Blackwell isn't shipping until Q4. Also Blackwell's lead will be short lived, because mi350x is coming out next year, and it will have a node and architecture advantage. So…
None of those things matter, if all you're looking to do is run your existing workloads on mi300x in the cloud. You get more bang per buck by going AMD.
ROCm has improved a lot. And you can rent mi300x in the cloud now. So if you have a program that runs on Nvidia GPUs, it takes no time to test it on a cloud mi300x. If it works you can use it and save some money in the…
I don't understand why they should use TensorRT. vLLM is much more popular and it was actually written for Nvidia. It also supports AMD hardware, so it's the appropriate tool to compare.
AMD doesn't have a higher PE ratio. You should use nonGAAP numbers because the GAAP numbers include Xilinx Googwill Amortization which skew the PE. AMD's PE is ~55. Nvidia's PE is above 70.
People used the same argument when saying AMD would never beat Intel in CPUs. Intel has a lot of software engineers. Also these days AMD has a good number of software folks, thanks to the Xilinx acquisition and the…
For us hobbyists used 3090 or new 7900xtx seem to be the way. But even then you still need to build a machine with 3 or 4 of these GPUs to get enough VRAM to play with big models.
zendnn is for CPUs. Pytorch ROCm version doesn't require you to change a single line of code. So all the same semantics like `device = "cuda"` work the same when using AMD GPUs.
libraries if they are popular in general are expected to work with a wide variety of code. Supporting code smell is a feature.
The author also says that AMD doesn't have Tensor cores. No shit, Tensor core is an Nvidia trademark. AMD has matrix multiplication units in their CDNA GPUs.
As a hobbyist you should look into spot instances in cloud. Way cheaper and faster than having to invest in your own hardware.
AMD has been improving a lot. Also you're comparing 3090ti to a 2 generation older RVII how is that a fair comparison?
Yes you can. And AMD's Linux driver is much better.
AMD's driver is much better on Linux than Nvidia's. And ROCm HIP stuff isn't obscure. It's what's needed to bridge Nvidia's proprietary vendor lock ins. Nvidia is literally cancer in this space.
There is also an article about how they did submit a score from one of these Chinese "exaflop" systems, for a different benchmark and it turns out it can only achieve the claimed performance at half precision:…
Epyc 2 (Rome) will have up to 64 cores.
> I guess AM4 also means no real improvements on the PCIe lane count The new Ryzen 3000 CPUs support PCIe Gen4.. so while the number of lanes will remain the same, their bandwidth could be doubled. Just announced Navi…
But the GC has to do it in a thread safe way which involves locking/synchronization. Otherwise you get nasty race conditions.
It's only a space heater when it's being used. These CPUs consume very little power when idle. So if you have a task that can leverage 32 cores.. that task should in theory complete that much faster than on your 65W…
Particularly in case of Threadripper, they use cherry picked binned cores so they can actually hit even [slightly] higher clocks than their mainstream cousins. We don't know the Threadripper 2's clocks yet, but last gen…
It's not happening because it's hard to do. Extracting instruction level parallelism is difficult(which is how CPUs gain IPC) and then also the cores have to be wider with exponential complexity. ie. At the point you…
Let's not forget Raven Ridge (Ryzen Mobile).. It targets lower power and lower performance, but it's still 3x faster than Intel's current solutions when it comes to graphics.
It's important to note their ARM CPUs aren't anything to write home about. Mostly a slightly modified version of ARM's own reference designs.
Intel has paid Nvidia billions in the past to use Nvidia's tech at the expense of AMD.. https://newsroom.intel.com/news-releases/intel-announces-new... Since both Nvidia and Intel are monopolies in their respective…