Setting for madmax

Chipset drivers can usually be found on the download page of the motherboard on the manufacturers website.

Yeah it’s frustrating. I have had this with the original plotter where it took me ages to get it running right and in the mean time you see people with similar setups doing much better for no clear reason.

One possibility imo you can cross of the list:
Madmax settings, this aint the problem. Even without tweaking you should be hitting ballpark good times.

So OS, hardware combi, drivers.
Trying Ubuntu, not a bad idea. Who know it might just work. If it doesnt, at least you know its not a OS problem

1 Like

CAN I USE THIS FOR MY CONFIG

MY PC : Ryzen 9 -3900XT / msi-b550 tomahawk/ 32Gb-3600 Ram / 1tb EVME SSD

currently i am using Madmax 0.6
can u Suggest me Best settings

Hi
MY PC : Ryzen 9 -3900XT / 32Gb-3600 Ram / 1tb EVME SSD

currently i am using Madmax 0.6 with the below .

.\chia_plot.exe -n 20 -r 22 -p KEY -f KEY -t E:\plot\ -d G:\final
but its very slow… can you suggest me settings

The price for 128 GB RAM is not justified for just few more plots per day. I reach 55 plots per day on a 5900x with only one 1TB 980 pro and 32gb 3600 RAM. So I imagine the 5950X can easily do 60 with the same setup.

1 Like

you seem totally clueless Adam. Why would you need adapters? You already have two PCIe 4.0 M.2 slots for your nvme. It can’t get better than that.

1 Like

Be careful, this adapter has one PCIe 3.0 slot and one SATA slot, much slower. In general M.2 slots provided by the motherboard are to be preferred anyway and your motherboard has enough of them.

2 Likes

Still, you only have to pay for RAM once and you can just plot like forever.

I am using 4 of 32GB DDR4-2666 RAM. They have the same price as a single 2TB SSD.

2 Likes

Yes, i know and sorry about that, but the reason is that i have bought all of this and i really don’t have a clue to set it up correctly.

My board have 2 x nvme slots, shot i put the nvmes in there or connect then through pcie (where the GPU is normally connected) :blush:

Thanks again

directly to the board will always be faster. Especially if your mb has 2 m2 gen4 slots

I did a ram disk a few months ago, it was slower than NVME, I suspect because CPU is using the same memory controller; the NVME pciex4 allows division of work. I was just curious that’s why I made the test, most ppl who have done ram disks have found it to be 10% slower than NVME.

I’m getting 20 minutes per plot using mad-max, just using -r 16 no other parameters; I have a 32core amd, with 64gb ram on this plot machine; 590 mobo

Before mad-max I was seeing using chia-plot-create I was seeing 5hr/plot, stagger 4-5, about 16/day, now I see 72/day

Lastly, I run mad-max with ‘nice -n 5’ that way it doesn’t hog the system, it doesn’t effect elapsed time.

1 Like

BINGO this is a big problem there are two m.2’s, only one if nvme pcie4x16, same for the pcie slow only one is pcie4x16, this means u can only have two per system that are fast

Worst of all is that your NVME pcie4x16 ( say MP600 corsair ) will act crazy and fail if u plut it into the m.2 sata slot.
Directly to board is not faster, I run on a pcie-extension board ( $5 alibaba ) with my MP600 I’m seeing same speed as the one next to the CPU. ( i have two mp600’s on this test system )

590 mobo two m.2’s but only the one next to cpu is nvme4x16, only first slot near cpu is pcie4x16

1 Like

I would stick with the singles, I’m not sure what this means to have two; given the controller is setup for one.
There is a big diff between m.2 sata drives and nvme4x16, and nvme4x
LOTS of FUD out there, but these tandem extension boards look like another scam, IMHO

I don’t think its a good idea to mix, if u really wanted the 4th m.2 u could just get another $5 board; I would not mix m2. sata, and m2. nvme-4x16

1 Like

Oh thanks for the help :blush:

So you mean that i should connect one of the nvmes m.2 directly into the board and the other one into the pcie 4x16 (where the GPU used to be) and i will get better results ??

Also i have the same requirements you are talking about 64 gb ram
2x nmve m.2
X570 pro wifi
5950x cpu

BUT… I’m getting like 1 plot/ every hour ??? How is that possible, when you are getting in the half time (30 min)?

NO U can’t do this -t, and -2 must be different this is why you are slow

I run two NVME 4x16 on my system,

nice -n 5 mad-max (chia-plotter) -r 16 -t /dev/nmve1 -2 /dev/nvme2 -d /dev/hdd0 -p key -f key

Nothing else, I’m getting plots every 20 minutes.

One thing I don’t like about MAX is they swap -t & -2, on the chia plotter (chia plots create) the -t is the working nvme, and the -2 is storage buffer,
On max the -t is the storage, and the -2 is the worker,
What max trys to do is OUT-SMART the user, their idea is that -2 is the faster drive, but in my case -t is the faster, but they’re actually both same-same MP600’s

When you do -t /dev/nvme1 -2/dev/nvme1

Then your not offloading work, your just running in circles chasing your tail, so instead of 20 minutes for a plot, your talking 2 hours, and complete killing of your NVME.

*** **-t /dev/1 -2 /dev/1 ** DONT DO THIS … replying to above

NO U can’t do this -t, and -2 must be different this is why you are slow

I run two NVME 4x16 on my system,

nice -n 5 mad-max (chia-plotter) -r 16 -t /dev/nmve1 -2 /dev/nvme2 -p key -f key

( 20 minutes per plot completion, I do batches of 36 [-n 36] )

-t is the the working nvme
-2 is the temp storage before hdd working dir
-d is the permanent storage hdd where the final resting place of the plots
these two working drives MUST be different, they both MUST be SSD

the -t MUST be nvme-4x16 (MP600 or EVO980)

The original idea of -2 was to buffer the plot, as the writing speed to -d can be 20 minutes; in fact now that we can make a plot in 20 minutes, its even more important to have a SSD for -2, so that -t, can start working on the next plot, while -2 buffers the last plot to -d

Now with mad max, he seems to be doing the smaller files on -2, and the bigger files on -t; so its critical to have TWO nvme4x16’s when using madmax

Like I wrote above, this issue of TWO NVME 4x16’s is not easy, as the MOBO m.2 has just one, an the 1st slot of the pciex has the second, and that is it. So u must have the pcie4x16 card for the second nvme; note that these MP600’s are like $200 each, this is not a poor mans hobby.

NOTE I keep saying NVME4x16, because most of the NVME’s sold are ‘sata’, and not the fast good ones, the ONLY I can recommend are SAMSUNG EVO 980, and Corsair MP600.

1 Like

Okay, i think maybe i figured out what the problem is… I don’t have the MP600 or EVO980.

I have the 2x viper vpn100 m.2 2tb… And i use 1 of them as the temp 1 and the other as temp 2.

My setting are like this :

.\chia_plot -n -1 -r 16 -u 256 -t “F:” -2 “G:” -d “J:” -p “KEY” -f “KEY”

F: is one of the 2 tb. Nvme
G: is the other 2tb. Nvme

And J is my HDD drive …

I think that my nmve is slow… But on the package it says that i get like 3450 mb read and 3000 mb. write.


Or maybe i should do as you where telling before, connect only 1 of the nmve directly into the board and the other nvme should be connected to the first GPU slot. With this converter in the top in the picture :

Hello guys! I’d like to request your help setting up properly our plotters.

Current config:
2x Intel Xeon E5-2697A v4 (32 core - 64 logical)
128GB RAM
4x 2TB nvme SSD
And a 18TB HDD for the final

We should start 4 at the same time on the 4 SSD or start only 1 at a time?

1 Like

this is a great NVMe. 1800 TBW… it doesnt get better than that…

1 Like

Hi, I have similar config and also did few tests.

5950x / 64gb 3600mhz / 2tb corsair mp600 pro
.\chia_plot.exe -p key -f key -r 16 -u 512 -v 128 -t D:\chia-temp\ -d F:\chia-plots
Total plot creation time was 1517.56 sec (25.2926 min)

So approximately 56 plots a day.

Edit: When running more than one plot in queue, times are getting a bit worse.
First plot takes 25 mins and all next ones - 28 mins, due to copy to hdd process.
So 51 plots a day.

1 Like

nice, that is pretty amazing, , but i can see that you only have 1 temp…

I Have my 1 of my nvme m.2 as temp 1 and the other nvme m.2 as temp 2… is that ok, or should i change …