I have a old computer with Ryzen 7 1700 cpu that I built 5.5 years ago (early 2017). Now I can still upgrade to the latest AMD 5700x / 5800x using the same motherboard, the only thing I need to do is update the bios. The price of 5800x has dropped to us$260 and us$235 for 5700x.
2 Likes
Userbenchmark has been slammed by the PC building community for having a very bizarre weighting system that favours CPUs with faster single-threaded performance. eg Discussing UserBenchmark's Dodgy CPU Weighting Changes - YouTube
2 Likes
Thatās good to know. I didnāt know AMD could last that long on a motherboard. Maybe I will consider AMD, next go around for my next upgrade.
I will check this out.
@alexatkin @blazini36 @lhkjacky you all changed my mind, I just checked out Alexatkinās video about userbenchmarks and it does appear as though they skewed the results in favor of intel. I figured that the the benchmarks there would have been an accurate representation because it was taking in so much data from so many different users and averaging it. But, there are clear benchmarks that AMD should have won in and they gave intel the win. It actually skewed the results for me so much, that it made AMD appear to be the underdog consistentlyā¦ I just started referring to some other benchmarks as well, and AMD tends to run cooler at times than Intel and tends to be more energy efficient and even destroys intel in some areas. Well, thanks again @alexatkin for sending that videoā¦ It really does change thingsā¦ I canāt even recommend that site anymore or even consider using it for buildsā¦ It does appear as though the results are rigged in favor of intel. But now, I find it completely unreliable as a reference pointā¦ I will use multiple sources next timeā¦ I had no idea how much AMD has improved!! Itās actually pretty impressive. I often was referring back to userbenchmarks to check previous generations and comparing them to AMD and I didnt see how much it improved because userbenchmarks skewed the results.
Do bare in mind the latest AMD CPUs are about the come out very soon, so it might be worth waiting a bit to see how prices adjust and if its worth jumping on their new platform.
1 Like
Sounds good, thanks for the tip.
I would love to have it on Linux. The only reason I use my Mac is for photo post-processing. Linux is my daily driver (Video Editing, Gaming, and Programming).
1 Like
Hello everyone,
No promises on an exact timeline, but if youāre interested in helping out with testing or giving us more detailed feedback (i.e., which distros to target, what format to distribute in) please make sure youāre signed up for the beta testing program within the next few weeks.
8 Likes
I have a couple of purpose built VEAI systems that run Windows specifically for that purpose. I absolutely hate Windows so If I could get this systems on Linux like everything else I have.
Not sure how many Desktop Linux users there are out there but trust me, Linux users pay for software that works too lol. I buy my Steam games, and I renew my VEAI license. I think this will eventually pay off if not immediately.
Oh, this sounds exciting. Where is the sign-up?
Hey! I am on RHEL 8.6 and the only thing holding me back from going full linux is this one app. It is an irreplaceable staple in my pipeline (And should be for anyone making computer graphics in 2022)
Benchmarks do one thing: measure how fast the benchmark is done 
There is no āthis one speedā - it all depends on he workload. Different benchmarks favour differenrt workloads - and most Benchmarks discussed on youtube or forums are biased towards gaming. Some offer some coputational workloads (geekbench, compubench, etcā¦)ā¦ But inference engines (which is what we use here) are often very different in needs than what is used in a benchmarkā¦
Modern CPUs have so many different ways to calculate āstuffā and with the introduction of P and E Cores, the picture gets more diffusā¦ Amount of caches, Numa Nodes, number of cores, ability to make the code run in SMT, Instruction sets, differnt compilers, Smart Access memory , etcā¦ all of these have an impact on speed.
So the best benchmark for VEAI is - VEAIā¦
1 Like
II think it would be nice if the Linux version works as a remote server and can be accessed and controlled remotely from Windows in the case of a GUI. Iām so curious how it will turn out
1 Like
Hello everyone,
Just wanted to give a reminder that interested parties should find their way over to the beta group!
4 Likes
I would support it ā I would also like to see your windows product tested in VMWARE Workstation.
Are you running in native VMWARE or Windows Workstation running in a VMWARE on-top of Ubuntu?
Is the beta group full at the moment? I asked for Permission and got no response yet.
You have very good points. I think the real money maker potential here, though, is enhancing the desktop/workstation workflow with behind the scenes software. Offer the Mac/Windows desktop/workstation version as the main interface, and then leverage those linux cli/workflow backends as an add-on enhancement to the familiar desktop frontends to allow those with older hardware, or with multiple boxes to incrementally increase their processing power, and reduce their video enhancement render times as they can afford to. Introduce a hardware abstraction layer, much as compilers do, to make the add-on, network rendering software component able to utilize whichever linux/unix box is out there as a slave processor to the desktop master, and abstract CPU versus GPU such that the software would just see and use what total instruction speed was available. Or, designate one linux box to be the master for the render queue, and the desktop or master would be responsible for interacting with Topaz as far as verification of subscription status, and video model refreshing/pulling. It would speed video rendering, and make additional speed incrementally more affordable for the end user. Using the workstation as the computer which interacts with Topaz will allow the Linux queue software to be more streamlined, modular, and potentially require less effort to maintain. Simplistic, I know. But like you said, the software is already built on top of existing Linux tools for the large part, and this would allow a more modular approach to increasing the capability of users, which can be monetized.
