Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AMD Open Source Driver for Vulkan

github.com

Post reply on HN
Print view View original

101–110 of 181 posts

Re: AMD Open Source Driver for Vulkan

#101

Post by dragontamer »

alanaktion wrote: post #97
zanny wrote: post #82

A lot of people are talking along the lines of "oh AMD is nice but... Nvidia". No, in 2019 all AMD GPUs this decade support OpenGL through 4.5, support Vulkan, and still really don't have a great OpenCL situation (rocm is out of tree on every distro and only supprts parts of 2.0 still). For gaming though, theres no reason not to get an AMD GPU. They are at near performance parity with Nvidia relative to their Windows…

I love AMD's innovation in this space, but for high-end gaming Nvidia is still destroying them in raw performance. RTX and G-Sync are definitely stupid, though they are adding limited Freesync compatibility to recent cards now. If AMD made something that'd beat my 1080 Ti for a reasonable price, I'd definitely buy it. I certainly don't like Nvidia's Linux drivers, but the majority of my non-IGPU needs are Windows-bas…

Well, the Radeon VII looks like it is around the 1080 Ti / 2080 for $699.

I think the main issue with AMD is that their compute drivers are clearly behind NVidia's. However, their ROCm development is now on Github, so we can publicly see releases and various development actions. AMD has been active on Github, so the drivers are clearly improving.

But I think it is surprising to see just how far behind they are. ROCm is rewriting OpenCL from scratch, HIP / HCC / etc. etc. is built on top of C++ AMP but otherwise seems to be built from scratch as well. As such, there are still major issues like "ROCm / OpenCL doesn't work with Blender 2.79 yet".

And since ROCm / OpenCL is a different compiler, it has different performance characteristics compared to AMDGPU-PRO (the old OpenCL compiler). So code that worked quickly on AMDGPU-PRO (ex: LuxRender) may work slowly on ROCm / OpenCL (or worst case: not at all, due to compiler errors or whatnot).

EDIT: And the documentation... NVidia offers extremely good documentation. Not only a complete CUDA guide, but a "performance" guide, documented latencies on various instructions (not like Agner Fog level, but useful to understand which instructions are faster than others), etc. etc. AMD used to have an "OpenCL Optimization Guide" with similar information, but it hasn't been updated since the 7970.

EDIT: AMD's Vega ISA documentation is lovely though. But its a bit too low level, and while it gives a great idea of how the GPU executes at an assembly level, it doesn't really have much about how OpenCL relates to it, or optimization tips for that matter. There are certainly nifty features, like DPP, or ds_permute instructions which probably can be used in a Bitonic Sort or something, but there's almost no "OpenCL-level" guide to how to use those instructions. (aside from: https://gpuopen.com/amd-gcn-assembly-cross-lane-operations/. That's basically the best you've got)

That's just the reality of the situation right now for anyone looking into AMD Compute. I'm hopeful that the situation will change as AMD works on fixing bugs and developing (there have been a LOT of development items pushed to their Github repo in the past year). But there's just so much software to be written to have AMD catch up to NVidia. Not just code, but also documentation of their GPUs.

Re: AMD Open Source Driver for Vulkan

#102

Post by llukas »

heroprotagonist wrote: post #70
tatref wrote: post #33

Earlier quoted context omitted.

I supported AMD several years because of this... Then I got tired of the low quality of both the open source, and the closed driver. Crashes, glitches in movies, flickering... Some time ago, I bought an Nvidia. It works like charm with the closed driver on Linux and windows. I do mainly games on Linux/windows, some gpgpu (machine learning with tensor flow), and the usual stuff. I couldn't be happier... Except if it w…

I thought it was interesting how they began forcing Windows users through a login page to associate identity, and a continuous background telemetry service, but not Linux users. The telemetry service needs to be running just to check for updates to the gaming card drivers. At least, I personally didn't have that experience on Linux with proprietary driver when the Windows gaming rig did. They likely determined it was…

You need to login for GeForce Experience not for driver. GFE is optional.

Re: AMD Open Source Driver for Vulkan

#103

Post by Rychard »

dragontamer wrote: post #101
alanaktion wrote: post #97

Earlier quoted context omitted.

I love AMD's innovation in this space, but for high-end gaming Nvidia is still destroying them in raw performance. RTX and G-Sync are definitely stupid, though they are adding limited Freesync compatibility to recent cards now. If AMD made something that'd beat my 1080 Ti for a reasonable price, I'd definitely buy it. I certainly don't like Nvidia's Linux drivers, but the majority of my non-IGPU needs are Windows-bas…

Well, the Radeon VII looks like it is around the 1080 Ti / 2080 for $699. I think the main issue with AMD is that their compute drivers are clearly behind NVidia's. However, their ROCm development is now on Github, so we can publicly see releases and various development actions. AMD has been active on Github, so the drivers are clearly improving. But I think it is surprising to see just how far behind they are. ROCm…

I bought my 1080 TI just over a year ago (December 2017) from Newegg for $750. (Newegg item N82E16814126186)

I'm glad AMD is finally catching up, but a savings of only $51 an entire year later doesn't exactly sound like a particularly great deal to me.

Re: AMD Open Source Driver for Vulkan

#104

Post by dragontamer »

Rychard wrote: post #103
dragontamer wrote: post #101

Earlier quoted context omitted.

Well, the Radeon VII looks like it is around the 1080 Ti / 2080 for $699. I think the main issue with AMD is that their compute drivers are clearly behind NVidia's. However, their ROCm development is now on Github, so we can publicly see releases and various development actions. AMD has been active on Github, so the drivers are clearly improving. But I think it is surprising to see just how far behind they are. ROCm…

I bought my 1080 TI just over a year ago (December 2017) from Newegg for $750. (Newegg item N82E16814126186) I'm glad AMD is finally catching up, but a savings of only $51 an entire year later doesn't exactly sound like a particularly great deal to me.

Welcome to the end of Moore's Law. 7nm is as expensive as 14nm was. Sure, you gained double the density, but it costs twice as much to make. So you only get improved performance / watt. Cost per transistor stayed equal in this 7nm generation.

NVidia's 2080 (roughly equivalent to the 1080 Ti) is also $699 to $799, depending on which model you get. Its the nature of how the process nodes work now.

-----------

Rumor is that the lower-end of the market will get price/performance upgrades, as maybe small-7nm chips will have enough yield to actually give cost-savings. But that's a bit of "hopes and dreams" leaking in, as opposed to any hard data.

For now, it is clear that 300mm^2 7nm chips (like the Radeon VII) are going to be costly. Probably due to low yields, but its hard to know for sure. Note that Zen2 and Apple chips are all at around 100mm^2 or so (which seems to indicate that yields are fine for small chips... but even then, Apple's Phones definitely increased in price as they hit 7nm)

Re: AMD Open Source Driver for Vulkan

#105

Post by deelowe »

zanny wrote: post #82

A lot of people are talking along the lines of "oh AMD is nice but... Nvidia". No, in 2019 all AMD GPUs this decade support OpenGL through 4.5, support Vulkan, and still really don't have a great OpenCL situation (rocm is out of tree on every distro and only supprts parts of 2.0 still). For gaming though, theres no reason not to get an AMD GPU. They are at near performance parity with Nvidia relative to their Windows…

Whats wrong with rtx (other than it being a bleeding edge tech that isn't well supported yet)?

Re: AMD Open Source Driver for Vulkan

#106

Post by shmerl »

bayindirh wrote: post #16
turblety wrote: post #3

So does anyone know now if AMD Vulkan GPU's are fully open sourced? i.e. can we build everything from source, firmware, drivers, app and then use it without having to trust any blobs?

AMD beta tester here. AMD separated HDCP and DRM related silicon from video acceleration units some time ago to be able to open source their GPUs completely sans the NDA bound stuff. Even this is a very big generosity and step from them for the Linux community. I'm sure that the firmware contains some highly proprietary and revealing information about some of their secret sauce. So, they won't be able to do it even i…

There is no secret sauce in the firmware, but it has HDCP garbage (which is causing it to remain a blob). AMD could provide one without HDCP as an open option, but I suppose they didn't see enough demand for it.

Re: AMD Open Source Driver for Vulkan

#107

Post by zepearl »

RussianCow wrote: post #71
zepearl wrote: post #25

Earlier quoted context omitted.

I did have an AMD card many (~10?) years ago, but I ended up replacing it with an nVidia because A) it was very noisy and B) Linux support was bad => I always used nVidia since then and never had problems with many types of setups and hardware, therefore I never switched back. I will reevaluate both brands for the next switch. I would like to use AMD because of its friendliness/efforts towards open-source, but it has…

Loudness has more to do with each individual card and how the manufacturer built/designed it than it does with AMD/Nvidia. Read GPU reviews online and you will see that, even within the same model (e.g. RX 580), some cards are known for being unreasonably loud while others are near silent.

You're fuzzy/vague and you assume that I did not perform a selection of what I decided to buy at that time and you state about huge differences of loudness between brands of the same chip without posting proof.

Re: AMD Open Source Driver for Vulkan

#108

Post by xvilka »

vorpalhex wrote: post #4

I'm glad AMD has consistently put in work to keep their drivers available to the Linux community, even if sometimes it's been less than perfect. I really hope that Nvidia eventually also open sources it's drivers.

When hell freezes over. Their management seems stuck to agressive anti-FOSS stance forever. Shame to every developer who works for them - it is like working for the evil.

Re: AMD Open Source Driver for Vulkan

#109

Post by blihp »

shmerl wrote: post #86
mrweasel wrote: post #9

Sadly I constantly hear people say that you should get an Nvidia card for both Linux and FreeBSD, because the drivers are better. While I'm sure that Nvidias driver a good, it's kinda sad that the attitude is that AMD is a better friend of the open source community, but yeah, we're going with Nvidia. I get why, you have stuff to do and Nvidia performs better, but still it a little annoying. OpenBSD seems to be the on…

> Sadly I constantly hear people say that you should get an Nvidia card for both Linux and FreeBSD, because the drivers are better. Not in the recent years. The trend has been positive for AMD and negative for Nvidia at least among Linux users for quite some time already. See: https://www.gamingonlinux.com/index.php?module=statistics&vi...

Not in all use cases (yet)... they still don't have a very compelling GPU compute story on Linux. There are reasons they've only made a minor dent so far and compute is one of them. As soon as they do, I'll be happy to switch my systems as I do like the direction AMD has been going the last couple of years on the driver front but they still have more work to do before it's viable for me.

Re: AMD Open Source Driver for Vulkan

#110

Post by shmerl »

blihp wrote: post #109
shmerl wrote: post #86

Earlier quoted context omitted.

> Sadly I constantly hear people say that you should get an Nvidia card for both Linux and FreeBSD, because the drivers are better. Not in the recent years. The trend has been positive for AMD and negative for Nvidia at least among Linux users for quite some time already. See: https://www.gamingonlinux.com/index.php?module=statistics&vi...

Not in all use cases (yet)... they still don't have a very compelling GPU compute story on Linux. There are reasons they've only made a minor dent so far and compute is one of them. As soon as they do, I'll be happy to switch my systems as I do like the direction AMD has been going the last couple of years on the driver front but they still have more work to do before it's viable for me.

> they still don't have a very compelling GPU compute story on Linux.

OpenCL AMD story has been rather flaky until recently, but supposedly ROCm is the most complete and open option there is now. You can also use Vulkan for compute purposes as far as I know. And Vulkan support isn't behind Nvidia in any way.

Some further Vulkan+OpenCL interop is still planned by Khronos.

See https://github.com/KhronosGroup/Vulkan-Ecosystem/issues/42#i...

Post reply on HN