I think this channel would gain enormously in popularity if short videos (10 - 20 mins) were offered.

@DigitalSpaceport

9 ай бұрын

Have done in past almost exclusively that, results were the opposite of gaining. When transitioned to longer numbers have gone up substantially. Have added timestamps to all long videos to assist folks better, I know they are long.

This is so cool!

Good 👍

I'm using the 16gb variant, bladebit cuda in linux on the newest chia update. System has 24gb ram and a 3060 non-ti. One proc I think it's a 6 core xeon v2. I'm currently plotting C7 and farming in same machine, the plot time is slow as hell like over an hour per but really for me plot times are not such a big deal. I do have 192gb ram coming for it (6x32) to fill the machine, I'll move over to the 128gb setting and we'll see how the times are. I think I'm currently at 40tb I'll slowly replot everything at C7 bladebit and maybe get another 8tb drive just because. This T320 server was actually harvested from the trash so nothing into it, I'm trying to fund all chia growth out of chia directly, it's a slow process!

Any word yet when 128gb bladebit will be on the GUI?

Nice video. Can you do a video about the cpu or GPU requirements for farming the compressed plots? Thanks

@DigitalSpaceport

9 ай бұрын

Ill include that section in next plotting farming video.

Another excellent video, i just dont want to update to win 11

@DigitalSpaceport

10 ай бұрын

I 💯 agree with you.l on that

i was wondering .since you need 8gb of vram to gpu plot in windows. can it also be done with sli and 2x 4gb vram nvivia gpu's ? got 2x gtx 680

Can you use multiple or select the GPU when there are multiple on the system? Also, just want to give you a huge THANK YOU for this tutorial is a massive help. Thanks,

@DigitalSpaceport

9 ай бұрын

Yes there is a GPU ID selection in the plotting tab that you can set to whatever number you want to use. 0 is the first GPU in your system typically.

@EfeArda1991

9 ай бұрын

What about farming compressed plots using a GPU? Can you chose what GPU you farm on? @@DigitalSpaceport

@DigitalSpaceport

9 ай бұрын

Yes the section in settings > harvester has a select GPU option. Specify number there.

I'm new to this and after chatting on FB Chia group.. i've learned I should DL the .db from Chia's site to get it setup on syncing faster than going through the software. I stopped syncing after it got to 462,466. It's almost 90% downloaded so far. So I run win 10, 3060TI, Ryzen 9-5950 and 16gb 2400 RAM. Should I upgrade with more RAM or get all new 128gb 3300..? I currently use my PC for Nicehash cpu\gpu mining. I wonder how it will run when farming too. ps. I subscribed. Oh.. I have 5 other hard drives in this pc. Just over 5gb total.

I feel at some point GH will start to lower fees to slow adoption of CBB plots. I have 16,000 GH plots. Going to wait a while longer before reploting anything. Thanks for all your informative videos!

@DigitalSpaceport

10 ай бұрын

I am waiting to see also. The curve isn't that big on the differences so max might, but then again he might not. The composition of addresses paying fees I assume skews larger farmers pretty heavy

@aguaponics

10 ай бұрын

@@DigitalSpaceport If Space farmers can keep a running total on us individually in our dashboard, I am sure they could post a daily total of what Max received from SF as a whole. At that point extrapolation could be figured as a whole block wide.

@DigitalSpaceport

9 ай бұрын

This is the kind of analysis pools should be offering TBH @@aguaponics

Hoping you can point me in the correct direction. I have a R720 256GB 3060TI running Windows 11. I was getting 10.5 minute plots which wasn't great but was happy with. A few days ago, that jumped to 30 minutes per plot. I have been plotting and have my main farm node on the R720. I stopped farming to see if that was the issue and no change. I then replaced my NVME with the same model....and no change. Any thoughts on what to look at? FYI...love your videos.

@DigitalSpaceport

8 ай бұрын

I'm assuming your just using a single socket? Check that windows 11 didn't autoupdate the driver and downgrade it. Mine did that like a few weeks ago. Annoying.

@pjdunc1956

8 ай бұрын

@@DigitalSpaceport By single socket do you mean 1 processor? I am running 2 as it is required for using a GPU.

@DigitalSpaceport

8 ай бұрын

Okay so your running pro version then. Did you check the installed driver version?

@pjdunc1956

8 ай бұрын

@@DigitalSpaceport Running the latest non gaming version of the driver. I have seen some postings that 2.1.X with Windows may be causing this problem.

For the Windows 128GB mode, how much "free" memory is required? While I have 128GB installed, it's about 25% used by Windows and other programs, leaving only about 95GB free unallocated.

@DigitalSpaceport

10 ай бұрын

If you have 128 installed, your good. It never crosses 127. Windows can "shrink" it OS footprint some and I'm sure Harold tuned the memory impact to be in a safe range.

@cornclose

10 ай бұрын

Great video and as is often the case, watched whilst insomnia is rife at 3AM (I mean that in a good way!), but was going to ask the same question and suggest a test with a 128GB system would be better than testing with 256GB. The Windows overhead is not insignificant and if one is to plot as well as farm on a 128GB system, we don’t want the whole thing grinding to a slow crawl with Windows thrashing the page file all the time. Can you whip out a few sticks of RAM and test with 128GB ? The results might help many, like me, on whether to stick with my current full node PC (which only has 32GB) before I upgrade to 128GB, or whether to not bother and source a workstation which supports more RAM. Thanks!

@DigitalSpaceport

9 ай бұрын

@@cornclose Okay did remove sticks and test 128GB and no change. However the real question here may well be around that 16GB mode. I observed like 22GB being used while ploting with it. I think you should test the 16GB mode but you for sure need good storage as it will get abused pretty hard. The Phison NVMe's are paid for and should arrive around end of this month and be the best "Cheap" option around but if you have an enterprise NVMe or a raid0 of several SSDs that might be OKAY. SLC is just so hard to beat.

@cornclose

9 ай бұрын

@@DigitalSpaceportwill the Phison SSD appear on your store when available ?

@smartsashko

9 ай бұрын

Nice video! Please advice me, im using 128gb ram, 3070ti and 2 nvme 3.0 512gb both. Looks like nothing from equipment load to 100 percent but gpu loading only 40-50 percents... Which component could be bottleneck?

Hi , how much disk space it will use for temp file @ 16GB ram mode ? thank you very much . I think maybe my intel 900P can work again .

@Small6300

10 ай бұрын

My 900p ,it's only 280G :(

@DigitalSpaceport

10 ай бұрын

Give it a shot. It's close. I don't have a firm number on the max write size needed but I was using a 320G drive and it fit.

Wow!! Thanks for the tuto. How can I plot with 6Gb GPU (RTX 2060) ?

@DigitalSpaceport

9 ай бұрын

I dont think 6GB works for plotting but you should test out Gigahorse and Chia GPU plotters to verify that. I dont have a 6GB GPU either to check against so if you could report back if you get either of those to work, that would be great. Use the highest C level as it will use the least amount of GPU RAM, so either a C7 - Chia or a C8 Gigahorse to test. Good Luck!

can you do a vid just on what you have in the notepad trying to do this but i cant find anything on the stuff you got on your notepad

@DigitalSpaceport

9 ай бұрын

Oh that's a good idea. Putting the text on screen for folks to read along with. That doesn't sound too technically difficult.

I'm guessing you can't run two plots at once? Then use the two nvme with each as single mode and final destination to a third drive with the moving script? Although I guess using 256 mode probably has a higher plots per day than dual 128 mode with nvme temp drives.

@DigitalSpaceport

9 ай бұрын

Not dual plotting really with GPU unless you have some seriously insane RAM and hardware running proxmox likely

@cracklingice

9 ай бұрын

@@DigitalSpaceport Yeah - it was just an interesting thought of 'I wonder if two 128GB modes does more plots per day than a single 256GB mode'. I haven't touched plotting since plots per day was the metric not how fast a single one could be created. Heck, I'm not even sure if RAM plotting outside of making a RAMdisk for temp was a thing back then. haha

@DigitalSpaceport

9 ай бұрын

Catch the latest video for me using full ram plotting with ramdisk cache temp for holding plots. Results were not as expected!

@DigitalSpaceport hi where can I grab power from for a 3060ti, is there a specific place or can I use a splitter cable at the motherboard power connector Dell T620

@DigitalSpaceport

9 ай бұрын

Oh I shot a video on this that I never released. There is a Power Addon Board. I did write about the T620 and have links to those specific PN in a written guide. Its old CPU plotting info but the HW section will be of use to you. digitalspaceport.com/dell-poweredge-t620-guide/

@vernonroffe1467

9 ай бұрын

@@DigitalSpaceport Great thanks the cable idea didn't work anyway, dell must have the pins wired differently. Had a look on ebay for the addon board but no joy none sold for months. On the dell forum apparently the board works from a T630 so have ordered one today also got to get the 1000w psu's, then hopefully ill be able to get plotting some Chia.

@DigitalSpaceport

9 ай бұрын

Dis the T630 board work? Is it in yet? Would like to add a note on this to the written article or something so others can share the knowledge

@vernonroffe1467

9 ай бұрын

hi @@DigitalSpaceport yes its in but I couldn't get the dell board specifically for the t620 anywhere but after some re-search on the net the one for the dell t630/640 model no X7C1K fits got it off eBay from China and fitted this morning the 3060ti appears to be working fine, the node is syncing at the moment so a day or so and hopefully get plotting. Its a 8 bay T620 so ill have a little breathing space next job getting the netapp 4246 up and running on the t620, what board and cable do i need for for that.

@DigitalSpaceport

9 ай бұрын

Thats great news! Those are much cheaper also vs the T620 sku number ones. Cable (for first JBOD daisy chain) -> geni.us/mCZCP and HBA Card -> geni.us/LSI-9207-8e and after the first cable from the HBA to the first disk shelf, you need just plain QSFP to QSFP cables -> geni.us/dK9Nk for adding additional DS4246 units to the first one. @@vernonroffe1467

Do you have one using ubuntu? I am having trouble getting it to read the cuda card m4000 i installed and could use a start to finish Ubuntu plotter setup

@DigitalSpaceport

10 ай бұрын

There is one shot some of the video got corrupted so it's a salvage operation but I think your bigger issue is Maxwell gen cards I don't think work. I think you need pascal or greater

@5minuteFlags

10 ай бұрын

@@DigitalSpaceport just circling back on this, i did manage to get the M4000 working it does have support for the CUDA 5.2 mentioned in the docs. I however neglected to connect the pci power cable. This and the correct drivers install seems to have me up and running. Now its time for some testing. Sorry to hear about your footage delete, thats got to be rough

@DigitalSpaceport

9 ай бұрын

I have been hit by power plug forgotten many times as well on GPUs lol @@5minuteFlags

How do I install bladebit in Windows?

Can this system work also withh just 128 GB RAM + a GTX 3060 with 12 GB GRAM or better wait for the official splitted RAM/SSD mode?

@DigitalSpaceport

10 ай бұрын

Yeah it can work with 128 + a 3060. This code is beta but if the plots farm and your signage points are passing you should be good. Easy to monitor that in chia GUI now. Just don't use 16GB windows plotting nor windows 10 and you should be good

Trying to split and use direct root (not folder) but it “Failed to create unsorted L disk buffer” Use D as a temp drive 1 and C as a temp drive2 (System win ssd is C). Cant understand why it fails. Any help please?

@DigitalSpaceport

10 ай бұрын

Are you spelling out --temp ? Also do you have enough space on the drives? This will push your C drives disk hard, ensure you have a backup of your OS.

@igortrandafilov2504

10 ай бұрын

@@DigitalSpaceport what do you call a “killer ssd” in your videos? I have 970pro and you advised it wont perform better than 9.5min which I have now. Can you please advise on ssd?

@DigitalSpaceport

9 ай бұрын

I ordered a batch of pSLC NVMe drives for chia plotting from Phison. They should be here around end of month and are super high BW and endurance for a cheap price. You could also use an enterprise NVMe as long as it has MLC/SLC and get good perf like this + endurance on it. @@igortrandafilov2504

Hi,I have the following configuration: Intel I9 CPU, NVIDIA GeForce RTX 4060 video card, 128 GB RAM, Temp1 and Temp2 SSD Silicon Power 2TB UD90 NVMe 4.0 Gen4 PCIe M.2 SSD R/W up to 5,000/4,800, I'm using Bladebit cudaplot with disk-128 mode , takes a long time about 2 hours to make a plot. Can you give me some help?

@DigitalSpaceport

4 ай бұрын

Try raid0 the 2 NVMe and lmk if that speeds it up

Can we get a “from scratch” start up for a chia farm using 2.1 and then creating compressed plots? Those of us who aren’t programmer lever farmers need some more simple language descriptions. Almost all videos out there now seem to assume your average user has a level of knowledge that most don’t. 95% (or more) of all computer users never even touch Linux for example, let alone know it even exists. I’m asking for a simple video that non-developer type people can actually use to keep up with the changing dynamics of plotting without getting left behind (I’m still stuck using OG plots for example because nothing I do seems to allow for showing the compression plotting options) Thanks

@DigitalSpaceport

8 ай бұрын

This is likely the guide you are looking for. It's a from scratch with the Chia official solution and windows. kzread.info/dash/bejne/fZWcvJRtoai9p9Y.html

@daviddejong779

8 ай бұрын

@@DigitalSpaceport 3 minutes into this video and it’s already language that your average user doesn’t understand. I really do appreciate the quick reply. People have a really good desktop, like a gaming desktop, with up to 128gb ram, windows 11 pro, a processor like a 10th gen or better intel. Can they create compressed plots by using the chia software, even if it means adding other software running in the background? I for example have a gaming computer currently that I’m using to farm about 300tib, but returns dropped a lot lately probably due to compression plots. I upgraded the ram to 128, have a 1tb m2 ssd, 10th gen intel, and 3060 (I also have a 3070ti I could put in if it helped). Is there still more needed just to keep up with compressed plotting? What made crypto mining catch on was that it became easy enough for everyday users to get into it. I was able to with chia so far, but unless I can just upgrade parts and the software running without getting into mods and scripts that I need to worry about constantly shifting, then chia is probably doomed to the same fate as any other software where the developers love it but can’t understand why it doesn’t catch on with mainstream. I am hoping I’m just missing something simple to start with compressing plots. I’d even be ok with the idea I need to wipe the system and start from scratch because maybe there is a software conflict in the background. Again, thanks for the quick reply. I’m hoping for a “here’s how to take a simple desktop and get started (if it’s possible) without a bunch of mods needed. Beyond appreciated

@DigitalSpaceport

8 ай бұрын

@daviddejong779 yeah the catch is GPU plotting is FAST. Faster than you can typically move to any single platter drive. There is almost no 1 size fits all that works for that every scenario, so you do end up needing either to plot direct to platter (12 min plots at least) or use a fast buffer and use some scripts. If you upgraded to 128GB ram, your good for Chia official plotting. Your specs could make a nice plotter/farmer combo I think. I don't know what else to tell you sorry. Yes your returns are down as you are not at as maximal a compression as there is. It's unfortunate I agree, but it is what it is for right now on the GPU front.

It is possible to use 32GB mode?

Is Win 11 mandatory? 😎 edit: Managed to create plot on Win10 Pro, but it's not visible in the Chia GUI... Anyways, I have 3060ti + 128gb 3200MHz, plot time 10 min, size 81.7 gb :D

@DigitalSpaceport

10 ай бұрын

I'm reading windows 10 has some issues. Do not plot in windows 10 for the time being.

@jarridgraham

10 ай бұрын

Same issue here Win 10. So may be a thing.

@jarridgraham

10 ай бұрын

I installed Win 11 still a no-go, found in the logs a mention of parallel_decompressor_count: 0 in the config, set that >0 and reload and then it showed. May have worked in Win 10 afterall.

@darker985

10 ай бұрын

@@DigitalSpaceport I installed Win11, I followed every step from your video and I''m getting this error (Fatal Error: VirtualAlloc failed.) Any ideas what might be? ;) Thx edit: managed to fix it, like Jarrid said, but I'm curious, your plots are 77.9, mine 81.gib8... is it because of GPU? ;)

@DigitalSpaceport

9 ай бұрын

W10 issue was invalid plots from what I read online, but the harvester may have worked. @@jarridgraham

for 16GB mode, it makes invalid plots on windows 10 & 11. check the plots.

@DigitalSpaceport

10 ай бұрын

This was addressed in the video as well and demonstrated, but the time to create a plot was fantastic for 16GB given the right HW.

@jungtony5221

10 ай бұрын

@@DigitalSpaceport and it also makes invalid plots on Ubuntu. I've tested 16GB mode on linux.

@DigitalSpaceport

10 ай бұрын

Mine worked in Linux interesting. What GPU and Ubuntu version are you on?

@DigitalSpaceport

9 ай бұрын

In case you had been using the GUI in Linux, they just fixed that in 2.0.1 creating invalid plots. Was just reading this. @@jungtony5221

I signed up for the channel 1.99 membership. Where do I find the 3% coupon code?

@DigitalSpaceport

9 ай бұрын

Welcome! On the channel, you should now see a Membership tab where the September code is linked from recent posts with stock updates (click 'read more' to find the 3% link and it should take you there). If you still can't find it, you can shoot me an email at social@digitalspaceport.com

@DigitalSpaceport

9 ай бұрын

Here's the direct link to the 3% off code. kzread.infoUgkxAuYc8lLZ2l263wFuT-wsmK9xagWYRren Let me know if you have any more issues getting this to work. Cheers.

My problem is windows 11 doesn't support xeon cpu so now I need to try the beta version.i have 256 gb ram

@DigitalSpaceport

10 ай бұрын

20:40 seconds into this video I show you how to get unsupported CPU and tpm bypass kzread.info/dash/bejne/pGhkj9SIktbVn9Y.htmlsi=nFxI029tGtpj3yKn

@jeffmcclure4047

10 ай бұрын

I got past the same issue with Rufus and a fresh win 11 iso install, you can bypass the checks for cpu and TPM, works fine

@jean-francoisaubin

10 ай бұрын

@@jeffmcclure4047 I will try later tonight thanks

Hi are these ram any good for my Dell t620 PECIFICATIONS PC3-12800r 1600Mhz DDR3 CAS-11 2Rx4 (Dual Rank) ECC REGISTERED 240 Pin 1.5v Module Capacity: 16GB Module Quantity: 2 Total Capacity: 32GB 12 Month Warranty. Just topping up to 128g Cheers 😊

@DigitalSpaceport

10 ай бұрын

That should be good 👍

@vernonroffe1467

10 ай бұрын

@@DigitalSpaceport hi where can I grab power from for a 3060ti, is there a specific place or can I use a splitter cable at the motherboard power connector Dell t630 ta

@vernonroffe1467

10 ай бұрын

T620 not t630

You on a rampage

Not sure what im doing wrong but im making the whole plot without issue but at the very end its says Warning: Thread did not exit properly. and the gui doesnt detect the plot

@DigitalSpaceport

10 ай бұрын

Which mode? 16 or 128m

@Bam222

10 ай бұрын

128@@DigitalSpaceport

@DigitalSpaceport

9 ай бұрын

Did you create the plot in Linux or Windows? @@Bam222

@Bam222

9 ай бұрын

@@DigitalSpaceport windows 11

@DigitalSpaceport

9 ай бұрын

@@Bam222 That one is new to me I would recommend you open a ticket on BB github.com/Chia-Network/bladebit/issues here and report with log added for them to asess issue. Are you plotting direct to a network share or something unusual? Describe your entire arrangement.

What driver version is required?

@DigitalSpaceport

9 ай бұрын

535

Its workiing on a 64gigs wiin machiines? 10k plots.. dont thiink iir worth to change GHC8 to BBC7

@DigitalSpaceport

10 ай бұрын

There is a Linux 64GB version but it's not made it's way to windows. The numbers still favor GH for earnings/space saved on the top end for sure. At lower C sizes it's less of a delta in space saved.

@sskula

9 ай бұрын

can you give us an example of your command line? my system with 64gb will not work and I think it is my command line. bladebit_cuda.exe -f -p -z 7 -n 1 cudaplot --disk-16 --temp1 d:\ --temp2 d:\Plots Kernel RAM required : 90240524288 bytes ( 86060.07 MiB or 84.04 GiB )

@sskula

9 ай бұрын

llocating buffers (this may take a few seconds)... Kernel RAM required : 90240524288 bytes ( 86060.07 MiB or 84.04 GiB ) Intermediate RAM required : 2999001088 bytes ( 2860.07 MiB or 2.79 GiB ) Host RAM required : 141733920768 bytes ( 135168.00 MiB or 132.00 GiB ) Total Host RAM required : 231974445056 bytes ( 221228.07 MiB or 216.04 GiB ) GPU RAM required : 6139441152 bytes ( 5855.03 MiB or 5.72 GiB ) Allocating buffers CUDA error: 2 (0x2 ) cudaErrorMemoryAllocation : out of memory *** Panic!!! *** Fatal Error: CUDA error cudaErrorMemoryAllocation : out of memory.

@sskula

9 ай бұрын

so I have bladebit V3 I think I need 3.1 beta 1 but I am not able to find it, does someone have a link for windows

Sir please tell me I have 128GB ddr4 ram Ryzen 5950x NVME subrent rocket 2 drive 4tb I want to buy 4080 16gb gpu I can crate c7 plot in 5 minutes on windows 11? Help me please

@DigitalSpaceport

10 ай бұрын

A 4080 will not perform better then a 3060ti for this plotting in split mode. Your NVMe is what really matters here and a single consumer drive, not in raid0, is going to max out around 10 mins.

@jakeleo1857

10 ай бұрын

@@DigitalSpaceport thanks sir

ok so where do you see farmer key and pool key?

@DigitalSpaceport

9 ай бұрын

Here is a companion video that shows the KEYS and more. Timestamped for you. kzread.info/dash/bejne/fZWcvJRtoai9p9Y.htmlsi=RXYJQM4YxHcny7xV&t=1105

@robertjerman3872

9 ай бұрын

So to make plots i need farmer public key and pool public key? Or something else? I made 10 plots and cant read them. Im totaly lost.i made them with windows 10. Or i need win 11?

@DigitalSpaceport

9 ай бұрын

You DO NOT want to use pool public key. Follow the video exactly. I show you the exact place to get the Plot Contract key in that timestamp. Windows 11 is better supported. Windows 10 some report makes bad plots.

Hello! Great video! Pls help me to find bottleneck od my system: i7 6950x 10c/20t 128gb 2400 ram oc to 2666 15 15 15 35 Raid 2 nvme 512 pcie 3.0 3070ti On windows im trying all scenarios and max plotting speed of compressed plots 12 mins:( Possible plotting very sensitive to PCIE 3.0? Because all my devices 3.0 only, no 4.0 on a motherboard Today will try with ubuntu but feel that speed will not increase) P.s. gpu loading only +- 50 percent

@DigitalSpaceport

9 ай бұрын

It's your NVMe. Even 2 consumer NVMe in raid0 will get hammered very hard by hybrid plotting mode. This is the reason I'm bringing in new Phison drives that have unlimited write performance vs having a cache that blows out.

@smartsashko

9 ай бұрын

@@DigitalSpaceport thanks for answer! Replaced with raid of intel ssd and now speed 8 mins. As i understand official support of compressed plots not working yet?

@DigitalSpaceport

9 ай бұрын

Oh good yes that sounds more like it. So yes it works and is official but with a lot of edges still being worked on.

Would there be any benefit to replotting madmax cuda plots with this?

@DigitalSpaceport

9 ай бұрын

not really. I am still GH and waiting for the new version that offers even more compression to match NoSSD

madmax still the king... faster and better compression

@wazuma

10 ай бұрын

But madmax comes with a fee

@cryptoandgaming

10 ай бұрын

@@wazuma seems worth it

Is this supposed to work on windows 10

@DigitalSpaceport

10 ай бұрын

Windows 10 is not producing valid plots in all combinations of system/gpu currently. You would have to wait for those fixes.

@jean-francoisaubin

10 ай бұрын

@@DigitalSpaceport if I plot on Linux will they run on windows

@DigitalSpaceport

10 ай бұрын

@@jean-francoisaubin yes plots are able to be used on any platform despite the platform they are created on. They are portable.

@jean-francoisaubin

10 ай бұрын

@@DigitalSpaceport will the Ubuntu app work on windows to plot or do I need run Ubuntu os

@DigitalSpaceport

10 ай бұрын

@jean-francoisaubin untested by me but I assume WSL2 will not work for this

windows 64 Gb works?

ok you did a great job of showing us what is possible, i hold no fault to you. But this is way to slow. 6 min is twice GH and lower compression. i am 59000+ plots at c8k32 on GH. i need chia to go to c9 or c10 to = GH c8k32 and even then it will take me a year to get it done. yea not moving to chia compressed any time soon. Why did you guys not just buy GH? It is clear you guys are out classed in this particular area, suck it up and buy GH for the sake of us miners. stop trying to reinvent the wheel. Pride comes before the fall.

I get 10x more time than your results with my 16GB of RAM & RTX 3070

@DigitalSpaceport

9 ай бұрын

That NVMe factors in HUGE here. The phison high endurance ones I used are ordered and on the way, ETA last week of Sept to ship out. An Enterprise NVMe that has SLC could also perform well for you or a large Raid0 of ssds but those all will cost a good deal more.

@This-Is-The-End

9 ай бұрын

@@DigitalSpaceport Clear explanation, many thanks.

@This-Is-The-End

9 ай бұрын

@@DigitalSpaceport PS: All my plots were bad plots on windows with 16GB Ram, they know about this issue on github, hope they will fix it soon.