# Time Warp Machine Learning **URL:** \ **Category:** Flame Questions\ **Tags:** machine-learning\ **Created:** [December 20, 2025, 7:16pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913 "2025-12-20T19:16:30Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![Hengy](https://avatars.discourse-cdn.com/v4/letter/h/96bed5/32.png) [@Hengy](https://forum.logik.tv/u/Hengy)\ **Post date:** [December 20, 2025, 7:16pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/1 "2025-12-20T19:16:30Z")
I’m doing some speed ramps on 4k material. Seems GPU is a no go. Using CPU instead. What is the gpu card you’re using to be able to render out 4K? ---
**Author:** ![ALan](https://avatars.discourse-cdn.com/v4/letter/a/a87d85/32.png) [@ALan](https://forum.logik.tv/u/ALan)\ **Post date:** [December 20, 2025, 8:24pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/2 "2025-12-20T20:24:29Z")
> **[NVIDIA RTX PRO 6000 Blackwell Workstation Edition](https://www.nvidia.com/en-us/products/workstations/professional-desktop-gpus/rtx-pro-6000/)** > > Experience Unparalleled AI and Graphics Performance. ---
**Author:** ![Hengy](https://avatars.discourse-cdn.com/v4/letter/h/96bed5/32.png) [@Hengy](https://forum.logik.tv/u/Hengy)\ **Post date:** [December 20, 2025, 9:13pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/3 "2025-12-20T21:13:38Z")
$8,000 will do ya. ---
**Author:** ![ALan](https://avatars.discourse-cdn.com/v4/letter/a/a87d85/32.png) [@ALan](https://forum.logik.tv/u/ALan)\ **Post date:** [December 20, 2025, 9:50pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/4 "2025-12-20T21:50:17Z")
That is the normal course of Flame…. when something is working in-efficiently, just throw more expensive hardware at it. ---
**Author:** ![ytf](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/ytf/32/109_2.png) [@ytf](https://forum.logik.tv/u/ytf)\ **Post date:** [December 20, 2025, 10:02pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/5 "2025-12-20T22:02:49Z")
The P6000 is not supported in 2027 and is only minimally supported in 2026.2. Save your Christmas money. ---
**Author:** ![randy](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/randy/32/6122_2.png) [@randy](https://forum.logik.tv/u/randy)\ **Post date:** [December 21, 2025, 2:35am UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/6 "2025-12-21T02:35:55Z")
You gotta provide a bare minimum of hardware details as far as what you’re working with. ---
**Author:** ![finnjaeger](https://avatars.discourse-cdn.com/v4/letter/f/e99b99/32.png) [@finnjaeger](https://forum.logik.tv/u/finnjaeger)\ **Post date:** [December 21, 2025, 9:44am UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/7 "2025-12-21T09:44:27Z")
that aint a p6000, nvidias naming is trash, its loke 4 generations newer. ---
**Author:** ![BenV](https://avatars.discourse-cdn.com/v4/letter/b/58956e/32.png) [@BenV](https://forum.logik.tv/u/BenV)\ **Post date:** [December 21, 2025, 1:04pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/8 "2025-12-21T13:04:52Z")
I was going to say get Twixtor but the ML feature has a bug on the Linux version so it’s disabled for the time being. Get a Mac, deliver and then return it… LOL ---
**Author:** ![randy](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/randy/32/6122_2.png) [@randy](https://forum.logik.tv/u/randy)\ **Post date:** [December 21, 2025, 1:06pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/9 "2025-12-21T13:06:46Z")
P6000 = Pascal = 2016 RTX 6000 = Turing = 2018 RTX A6000 = Ampere = 2020 RTX 6000 Ada = 2023 RTX 6000 Pro Blackwell = 2025 ---
**Author:** ![ytf](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/ytf/32/109_2.png) [@ytf](https://forum.logik.tv/u/ytf)\ **Post date:** [December 21, 2025, 1:55pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/10 "2025-12-21T13:55:00Z")
?. The P6000 pascal is EOL. Not sure what the name has to do with it. ---
**Author:** ![finnjaeger](https://avatars.discourse-cdn.com/v4/letter/f/e99b99/32.png) [@finnjaeger](https://forum.logik.tv/u/finnjaeger)\ **Post date:** [December 21, 2025, 2:05pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/11 "2025-12-21T14:05:54Z")
i thought you commented on alans link maybe not 🤷 ---
**Author:** ![randy](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/randy/32/6122_2.png) [@randy](https://forum.logik.tv/u/randy)\ **Post date:** [December 21, 2025, 2:11pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/12 "2025-12-21T14:11:37Z")
Finn and I both thought that you wrote P6000 is not supported replying to Alan, and we assumed you thought P6000 was the same as the 6000 Pro, which, makes total sense, but is of course incorrect. ![nothing GIF](https://us1.discourse-cdn.com/flex020/uploads/thedepartmentofexternalservices/original/2X/5/5baf67eaedafb3ec5458aa1809011a9b996f2c67.gif) ---
**Author:** ![ytf](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/ytf/32/109_2.png) [@ytf](https://forum.logik.tv/u/ytf)\ **Post date:** [December 21, 2025, 2:21pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/13 "2025-12-21T14:21:50Z")
If I had done that I would have used the little reply button on his comment. Not sure if you’ve noticed, but when you do that, there’s a little arrow pointing to the member’s icon in the upper right of the reply. But seriously, do you think I’m that dumb? Aside from the fact that I know the difference between a P6000 and an RTX 6000 pro, I’m not nearly stupid enough to challenge Alan on techie shit. ---
**Author:** ![finnjaeger](https://avatars.discourse-cdn.com/v4/letter/f/e99b99/32.png) [@finnjaeger](https://forum.logik.tv/u/finnjaeger)\ **Post date:** [December 21, 2025, 2:59pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/14 "2025-12-21T14:59:33Z")
But the where did the P6000 comment come from, i guess that lead to our confusion 🤣 nobody said anything about a p6000 ---
**Author:** ![ytf](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/ytf/32/109_2.png) [@ytf](https://forum.logik.tv/u/ytf)\ **Post date:** [December 21, 2025, 3:55pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/15 "2025-12-21T15:55:40Z")
It’s pertinent to the whole thread. It’s why the OP can’t do ML speed ramps on the GPU. Also the fact that Alan mentioned the latest recommend as well as the fact that we need to continually upgrade our hardware to keep up with Discreet. I thought it a good time to re-enforce the fact that all of us who still have the P6000 need to make an investment if we want to upgrade to 2027 in April. ---
**Author:** ![Hengy](https://avatars.discourse-cdn.com/v4/letter/h/96bed5/32.png) [@Hengy](https://forum.logik.tv/u/Hengy)\ **Post date:** [January 13, 2026, 8:01pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/16 "2026-01-13T20:01:56Z")
Are people still using [talosh](https://forum.logik.tv/u/talosh)’s time warp? Didn’t need much additional GPU memory. Wish I had it now. ---
**Author:** ![Sinan](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/sinan/32/155_2.png) [@Sinan](https://forum.logik.tv/u/Sinan)\ **Post date:** [January 13, 2026, 8:09pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/17 "2026-01-13T20:09:39Z")
You can switch to CPU mode if you don’t have a lot of VRam. It will crawl but get there. ---
**Author:** ![kily](https://avatars.discourse-cdn.com/v4/letter/k/a88e4f/32.png) [@kily](https://forum.logik.tv/u/kily)\ **Post date:** [January 13, 2026, 8:54pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/18 "2026-01-13T20:54:29Z")
Here. There is a tip with talosh’s TWML. If you run out of vmemory, and you don’t need much additional, you can set the clip’s tw, exit flame, decrease the vram reserved for flame in flame setup app, lets say 15%, start flame, (maybe flame run slow and laggy) and launch the hook. It will run with more Vram free, and probably you can render the tw. Restart flame reseting the original value of vram reserved in flame setup. ---
**Author:** ![pixelmixer](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/pixelmixer/32/2994_2.png) [@pixelmixer](https://forum.logik.tv/u/pixelmixer)\ **Post date:** [January 13, 2026, 10:38pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/19 "2026-01-13T22:38:30Z")
How about a way to purge VRAM whilst in the app? ---
**Author:** ![kily](https://avatars.discourse-cdn.com/v4/letter/k/a88e4f/32.png) [@kily](https://forum.logik.tv/u/kily)\ **Post date:** [January 13, 2026, 10:53pm UTC](https://forum.logik.tv/t/time-warp-machine-learning/13913/20 "2026-01-13T22:53:56Z")
The point is flame uses a portion of vram room itself, no matter if you purge vram. Reducing that by starting flame with this tip, you can claw back a bit of VRAM, just like Hengy says he needs. It’s the same for me. Doing this allows me to use TWML for 4K. [Next page](https://forum.logik.tv/t/time-warp-machine-learning/13913.md?page=2)