Editorial AMD is Becoming a Software Company. Here's the Plan

[OolongGoat]

This honestly feels like something that should have started the moment AMD acquired ATI. And then they wasted that momentum on weird initiatives like pushing APUs as the “future of heterogeneous computing” and making GPUs with massive compute potential (higher than that of NVIDIA at the time), but with absolutely no software stack to even support it. It’s good that they finally woken up to the reality that apart from console chips the Radeon division was essentially a dead weight stuck in limbo for a decade now, but they have to actually commit to transforming themselves this time around and prepare contingencies other than banking heavily on the AI fad.

I was thinking the same thing after reading this. I remember seeing all the Fuzion is Future marketing and there were big promises made then as well. I really hope they do make a turnaround in this venture. They have always had impressive hardware specs on paper, but the implementation has seemed to always miss the mark. The hardware isn't as good if there isn't the software ecosystem around to support it.

The last AMD products I owned were the Fury X/Nano. I have tried some of their newer GPUs, but I just haven't had the best luck with them. When using DVDFab for video work, it seems that CUDA is better optimized than AMD APP.

Their All In Wonder products were great back in the day, the software seemed to function really well. I am surprised they haven't done something like an AppleTV or a Nvidia Shield product.

 

[Gooigi's Ex]

Not gonna lie, I busted out laughing when I read the title because AMD becoming a software company? Wish them the best of luck because they gonna need it.

The quality of their software has much improved over the years and damn near on par. However, they are far behind when it comes to features and capabilities. Probably behind Intel when it comes to software.

 

[hsew]

With all the big customers are becoming their own semiconductor engineers, I’d say this ship has sailed… this would have been the move to make 10 years ago at the latest.

 

[A Computer Guy]

Wasn't already a software company? They did some programs and all failed along the way. The AMD Radeon ramdisk and the the AMD Radeon ProRender come to mind.

The ram disk is still sold to unlock the default limits. I'm using it currently. Too bad they don't provide it as a bonus with each CPU sold.

 

[AusWolf]

What a bummer...

I guess the days of hardware-agnostic software are coming to an end.

 

[mb194dc]

My suggestion, if AMD want to match Nvidia commercially, is that they need to up the bullshit factor, create some features Nvidia doesn't have. All the better if there's actually a practical advantage to them, that doesn't really matter... Or it hasn't for Nvidia anyway. AMD products are good largely. Their marketing and brand image is nowhere near as good as Nvidia though. Even if the stock trades at 200+ P/E ratio somehow.

The problem is that Nvidia is always dictating the sales narrative on the front foot, be it from RTX, Upscaling with DLSS, through the Cyrpto Mining boom and on to AI GPUs for LLMs being our technological savior. A lot of what they're pushing is very overhyped, just used to up their margins and sell inferior product, or creating features that are largely useless but give them an advantage in some way. Few reviewers are prepared to call a spade a spade when they release something like the 4060 series.

 

[A Computer Guy]

I hope this means they will fix Ryzen Master and ensure high quality UEFI.

 

[Ruru]

They should not have mentioned Tesla here.

My bet? They will flop hard, it is too late.

AI? We all know it will not take off. It is a marketing tool, nothing else.

This, this and this. Feels already that AI is a more common in marketing than gaming.

 

[lilhasselhoffer]

So...does anyone else remember that wonderful time in the past where hardware was constantly moving forward, and Moore's law was easy to harness as a cheap way to have shortcomings disappear beneath a pile of raw computational horsepower?

Does anybody else remember when everything started to slow down? Yes, the Intel shenanigans with 4 core chips for a decade was entirely because they didn't have AMD to compete against, but once that was sorted we got back to generational improvements not of 30-50% (read: peak in the 90's and early 00's), but of 7-13%. Anybody else remember when the concept of a generational refresh was something absolutely silly, because nobody would buy another $150 card if it basically only showed minor improvements. Yeah, I am old enough to remember all of that and team Blue, Red, and Green all claiming that we were reaching near peak hardware performance with generational improvements basically being decreased to minor changes.


So what are all three of the teams doing? Well, those old enough should recognize this as a netburst moment. What I mean by that is that instead of pursuing an ever less possible improvement in performance you can write entirely new rules. Just like Musk claims that Tesla should be viewed as a software company...(snicker)...if AMD, Intel, and Nvidia all become software purveyors they don't have to live and die on improvements in tangible things. They can claim something like a 30% increase in performance in their latest release, and in reality just have code optimization cover the entirety of their flaws to get that magical performance boost.

Let me offer a parallel. Warframe is a game made by Digital Extremes. Through decades of spaghetti code, at one point they were supporting all the way back to 32 bit windows xp. Kinda nuts, but one day they decided they needed modern instruction sets...and stopped supporting basically everything before Sandybridge and Bulldozer. This allowed them to cull the code base, and increase performance...but realistically it was improving the performance to a piece of software not bogged down with a decades old restriction set.


What truly frightens me though is the future of FPGAs and mismarketing. Let me state that Nvidia has specific processing for ray tracing....right? You've also got AI accelerators. Then you can design a bunch of other unique things...and claim how special you are. If you want to do this on a budget you get a big old FPGA and keep half a dozen images for accelerators...flashing them as needed to make the ultimate processor and co-processor combination. This then gets down to literally being software companies running relatively commodity hardware...at which point instead of companies trying to improve performance we'll be back to who can stick the biggest FPGA on, or who has the one ASIC image to flash to your shiny new FPGA. I don't even want to read all of the AI buzz word garbage...because it by definition is a fad. If you believe that anybody is going to sell you a consumer viable priced system to effectively allow reasonably quick local instances of LLMs then you didn't read that recent article where China was buying H20's from Nvidia for $12,000 each.

 

[Assimilator]

This. Then again, NOT starting on it is a sure fire way to leave the building.

Yeah it's very much a damned if you do, damned if you don't scenario. My biggest fear is that 2029 rolls around, AMD has finally got some decent software, the AI bubble bursts, all the software engineers get let go and the software gets outsourced, and the company squanders the goodwill they managed to build in that intervening half-decade. That would completely destroy any future possibility of getting into businesses, and probably end AMD as a company.

tl;dr right decision, wrong reason, way wrong timing.

 

[evernessince]

I don't think AMD really had a choice but to make this move. Nvidia's proprietary lock-in has continued to extend their domination of the market. It's impossible for AMD to be only be a hardware company when their competitor is increasingly vertically integrating into the market all the way up and down the software and hardware stack. We have Nvidia tech in most games nowadays and even mice, monitors, ect.

One has to hope that ROCm was part of this initiative and that they actually started buffing up their software hires at least a year back. Otherwise, as @Assimilator pointed out it takes years to build up the required expertise.

I hope that AMD continues to push open source because I frankly do not want even more features in games that can only be used on a specific video card brand.

 

[mtosev]

Interesting choice by amd.

 

[Deleted member 223428]

I'd like to think this is all part of a long term strategy, carefully planned and executed, with roadmaps and solid goals.

The realist in me thinks the last board meeting discussed Nvidia's success, how they became the most valuable company on the planet, and the response was:

 

[AusWolf]

I hope that AMD continues to push open source because I frankly do not want even more features in games that can only be used on a specific video card brand.

This a million times over. Unfortunately, I read the opposite in articles like this one.

 

[dragontamer5788]

I guess the days of hardware-agnostic software are coming to an end.

Its more about the decline of Microsoft Windows.

Microsoft cares about trying to have software run on the most hardware as possible. Linux really doesn't (indeed: its hardware-devs that are pushing Linux so much).

Eventually, people will realize the problem and we will swing back to more portable systems (like Java, more portable OSes, etc. etc.). But for now, hardware specific is king.

 

[unwind-protect]

If you believe that anybody is going to sell you a consumer viable priced system to effectively allow reasonably quick local instances of LLMs then you didn't read that recent article where China was buying H20's from Nvidia for $12,000 each.

That is for training the models.

Consumers running the trained form of these LLMs locally do inference which is nicely done with cheaper hardware.

I hope that AMD continues to push open source because I frankly do not want even more features in games that can only be used on a specific video card brand.

I am optimistic because I think AMD has no choice. NVidia's customers hate to be locked in. I'm talking about customers like Google here. They want choice. They want the certainty that the software is not developed in an unwanted direction in the future. AMD better complies with that.

 

[qlum]

Good that they are making a bigger push for software, they have been slowly scaling there for a while, and just hiring a shitload of engineers is bound to lead to wastage, so in principle this timeline is not too bad, but especially the gpu department needs more engineers desparately, for about everything.

 

[Firedrops]

AMD: No rocm support for everything except the latest gen highest end GPUs

also AMD: Why don't anybody develop for AMD hardware?????????????????????????????????????????????????

They think the tinkerers are building for Nvidia on Titans/x090s? Most of us were working on 2-3 generations old x60 class GPUs until the Nvidia ecosystem became reliable enough to justify high end purchases.

 

[ir_cow]

If AMD can't make good drivers for their own GPUs, how can anyone trust them to be a software company?

 

[unwind-protect]

If AMD can't make good drivers for their own GPUs, how can anyone trust them to be a software company?

I think their recent-ish hiring indicates that they want to change that.

 

[Soul_]

AMD: Drops rocm support for everything except the latest gen highest end GPUs

also AMD: Why don't anybody develop for AMD hardware?????????????????????????????????????????????????

They think the tinkerers are building for Nvidia on Titans/x090s? Most of us were working on 2-3 generations old x60 class GPUs until the Nvidia ecosystem became reliable enough to justify high end purchases.

Precisely this. Software is where they absolutely have the biggest gap. Hell even the software they should be producing to support their hardware is not always the best and sometimes inconsistent.

They could be about 10 years behind at this point. A lot of catching up to do. Problem with this is, in the short to medium term they wouldn't be able to sell the AMD ecosystem concept that others have already built, and that takes time to build, not even talking about adoption at this point.

 

[eidairaman1]

My suggestion, if AMD want to match Nvidia commercially, is that they need to up the bullshit factor, create some features Nvidia doesn't have. All the better if there's actually a practical advantage to them, that doesn't really matter... Or it hasn't for Nvidia anyway. AMD products are good largely. Their marketing and brand image is nowhere near as good as Nvidia though. Even if the stock trades at 200+ P/E ratio somehow.

The problem is that Nvidia is always dictating the sales narrative on the front foot, be it from RTX, Upscaling with DLSS, through the Cyrpto Mining boom and on to AI GPUs for LLMs being our technological savior. A lot of what they're pushing is very overhyped, just used to up their margins and sell inferior product, or creating features that are largely useless but give them an advantage in some way. Few reviewers are prepared to call a spade a spade when they release something like the 4060 series.

The biggest issue with Advanced Micro Devices is their Lack of Advertising.

 

[OneMoar]

AMD Whos software is worst in class is becoming a Software company... but they have _ALWAYS_ Been a software company

 

[A Computer Guy]

With Windows becoming more intrusive maybe it's time for AMD to partner to produce a serious alternative desktop? Linux + Steam + AMD = The future of desktop gaming?

 

[unwind-protect]

With Windows becoming more intrusive maybe it's time for AMD to partner to produce a serious alternative desktop? Linux + Steam + AMD = The future of desktop gaming?

Their contributions to Linux and other open source software are quite impressive lately.

Wine gaming on Linux is quite steady in its progress.