SRBMiner-MULTI AMD & NVIDIA & INTEL & CPU miner

[dd341]

Pearlhash
2080ti went from 82TH to 63 TH overnight

P\s are there any chances for future Pascal support?
Seems like there are some miners already available but non are optimized or updated for past couple months and hash is 7-10TH for 1080.
Everybody who still has 1080\1080ti\P40\P102-100\p104-100 probably still searching for decent pearl miner...

[maXonja]

V3.5.1
+ Improved algorithm 'pearlhash' for all NVIDIA GPUs (except Turing architecture)

Nice, that's it

[MinerGuy82]

V3.5.1
+ Improved algorithm 'pearlhash' for all NVIDIA GPUs (except Turing architecture)

Now some hashrate recover, but 10% more powerdraw. Best Efficency form me on 3070 was with 3.4.8 version.

[1971ECPT]

This printscreen is after fork with SRB 3.5.1.
With yellow, I wrote what was BEFORE fork with SRB 3.4.8

https://i.postimg.cc/63HCPP80/SRB-3-3-1.jpg



Now, 4-5Watts more consumption for each card and less 4MH less for each card.

I think is acceptable.

[MinerGuy82]

This printscreen is after fork with SRB 3.5.1.
With yellow, I wrote what was BEFORE fork with SRB 3.4.8

https://i.postimg.cc/63HCPP80/SRB-3-3-1.jpg

https://i.postimg.cc/MMQcMdbM/SRB-3-3-1.jpg

Now, 4-5Watts more consumption for each card and less 4MH less for each card.

I think is acceptable.

Depend how many GPU you need.
Ive got More 100watt per Rig. In 24 Hours are 2,4kw more per Rig. But if this a a consequence of the need of the new version after 4.8 ok... and however with the 4.5.1 I recoverd only 5% of hashrate, not 10%.

[marihome]

I found a bug in version 3.5.1 on Windows 10. When starting up, only the first three cards can be activated successfully, even though there are five cards in total. It shows a “CFX initialization failed” error.

Version 3.5.0 works normally. I think increasing the virtual memory might solve the problem, but my hard drive only has a few dozen GB of free space, which is not enough.

I am only using PRL. Why does it show CFX? Is this a bug? Version 3.5.0 works fine, but version 5.0 has much lower hash rate.

[Metroid]

3060ti rig is showing strange power consumption on prl, at software level is showing 1660w, at wall is showing around 2000w, 3070 rig at software level is showing 1830w, at wall 2050w.

[dd341]

3060ti rig is showing strange power consumption on prl, at software level is showing 1660w, at wall is showing around 2000w, 3070 rig at software level is showing 1830w, at wall 2050w.

Oh, same here. Software shows 10amps Usage but it's 13amps at the wall. I didn't think that was miner issue before, but now that confirms that.
I can test again later, but i stopped using SRB because 2000 series hashrate is now A**

[Metroid]

3060ti rig is showing strange power consumption on prl, at software level is showing 1660w, at wall is showing around 2000w, 3070 rig at software level is showing 1830w, at wall 2050w.

Oh, same here. Software shows 10amps Usage but it's 13amps at the wall. I didn't think that was miner issue before, but now that confirms that.
I can test again later, but i stopped using SRB because 2000 series hashrate is now A**

usually at the wall power will always be more, in my case the 3070 is right, the 3060ti is not, on other versions i did not have this issue.

[dd341]

3060ti rig is showing strange power consumption on prl, at software level is showing 1660w, at wall is showing around 2000w, 3070 rig at software level is showing 1830w, at wall 2050w.

Oh, same here. Software shows 10amps Usage but it's 13amps at the wall. I didn't think that was miner issue before, but now that confirms that.
I can test again later, but i stopped using SRB because 2000 series hashrate is now A**

usually at the wall power will always be more, in my case the 3070 is right, the 3060ti is not, on other versions i did not have this issue.

Yes, but... Well Gminer shows 10amp usage and wall usage shows 11amp.
SRB 3amp difference is 660w.

[FP91G]

Why do you look at power consumption data in the program? It usually gives inaccurate values. I always measured the power consumption of the motherboard, processor, and hard drive separately from the wall, and then measured the power consumption of the entire mining rig. This way, I always saw the actual power consumption of the video card from the wall, assuming all the GPUs in the rig were identical.

[Maliki2]

Getting a weird error with the newest release when I go to mine Pearl with my 9060xt I get a No GPU devices found, y ou have to define at least 1 GPU with --GPU-id parameter. Any ideas?

[dmarijan]

Getting a weird error with the newest release when I go to mine Pearl with my 9060xt I get a No GPU devices found, y ou have to define at least 1 GPU with --GPU-id parameter. Any ideas?

Changelog from v3.5.0:

Removed AMD RDNA2 support for algorithm 'pearlhash'

[marihome]

SRBMiner-MULTI 3.5. 1
Pressto switch to the next poolto switch to the previous pool'ress0-9 to disable/enable GPU 0-9, shift+0-9 for GPU 10-19'ress
\lgorithm/sGpu miningpu mininglatchdog
pearlhashenableddisabled
L2.00% fee」
enabled
[2026-08-09 15:33:39] Hashrate represents the 60 second average value - max hashrate after few minutes
Pearl backend error E2012
Pearl backend error E2012
[2026-08-09 15:33:41] Pearl GPU3 init failed: Pearl backend error E1013
[2026-08-09 15:33:41] GPU3 couldn t setup ctx->algorithm
Pearl backend error E2012
Pearl backend error E2012
2026-08-09 15:33:41
15:33:4115:33:4115:33:4115:33:41
Pearl
GPU4 init failed: Pearl backend error E10l3
2026-08-09
GPU4
couldn t setup ctx->algorithm
pearlhash
[2026-08-09
GPUO
A0A0
[2026-08-09
GPU1
pearlhash
[2026-08-09
GPU2
Ipearlhash
[2026-08-09
133.45
Connected to 172. 17.0. 93:1200 [pearlhash]
ob received [00000000_2097152 ] block height 97435 [diff: 2097152] [pearlhash][0]
2026-08-09
133:46
15:34:05
Difficulty 2097152. 0000000000 [pearlhash][0]
2026-08-092026-08-09
Job received [00000001_2097152 ] block height 97435 [diff: 2097152] [pearlhash]

•                                                                                           3.5.0is ok but 3.5.1if wrong..

[doktor83]

V3.5.2
+ Improved algorithm 'pearlhash' for 3000, 4000 and 5000 series GPUs
+ Improved algorithm 'pearlhash' for CMP170HX, A100 GPUs
+ Improved algorithm 'pearlhash' for some 2000 series GPUs
+ '--oc-coffset-delayed' is now also applying core clock
+ Bug fixes

[egocentr]

V3.5.2
+ Improved algorithm 'pearlhash' for 3000, 4000 and 5000 series GPUs
+ Improved algorithm 'pearlhash' for CMP170HX, A100 GPUs
+ Improved algorithm 'pearlhash' for some 2000 series GPUs
+ '--oc-coffset-delayed' is now also applying core clock
+ Bug fixes

3-5-2. Hello. QTC has noticed dips in GPU consumption. The graph is jagged.

[doktor83]

V3.5.9
+ Improved algorithm 'pearlhash' for unlocked CMP 90HX (a lot)
+ Improved algorithm 'pearlhash' for 2000 series GPUs (both efficiency and hashrate)
+ Minor improvements on 'pearlhash' for 4000, 5000 series GPUs
+ Added optional 'pearlhash' kernel for 4070ti, 5070ti (use --pearl-k2)

V3.5.8
+ Added algorithm 'sha256d_csd' (Compute Substrate) for AMD[RDNA]/NVIDIA/INTEL GPUs, fee 1.0%*
+ Improved algorithms 'xelishashv3', 'verushash' on CPUs
+ Removed algorithms 'progpow_sero', 'progpow_telestai', 'walahash', 'karlsenhashv2' and related dual implementations

V3.5.7
+ Added algorithm 'randomdrgx' (DragonX) for CPU mining, fee 0.85%*
+ Added algorithm 'blake3_an' (Alphanumeric) for AMD[RDNA]/NVIDIA/INTEL GPUs, fee 2.0%*

V3.5.6
+ Added algorithm 'randompcn' (Pcoin) for CPU mining, fee 0.85%*
+ Added algorithm 'randomzqvx' (Zerqavon) for CPU mining, fee 0.85%*
+ Added support for SM70 (Volta) architecture on algorithm 'pearlhash' [tested on Tesla V100 SXM2]
+ Improved algorithm 'pearlhash' for NVIDIA 2000 series GPUs (~3.5%)

V3.5.5
+ Added algorithm 'sha3t' (Bitcoin3) for AMD/NVIDIA/INTEL GPUs, fee 1.0%*

V3.5.4
+ Improved algorithm 'pearlhash' for 4000 series GPUs (mostly efficiency)
+ Improved algorithm 'pearlhash' for H100 GPUs (for efficiency mode use --gpu-cclock 900)
+ Minor efficiency improvements for some 2000 and 3000 series GPUs

V3.5.3
+ Just another emergency hardfork for Pearl coin (algorithm 'pearlhash') - MANDATORY UPDATE
+ Added optional 'pearlhash' kernel for 3070, 3070ti which can be used with higher core clocks (use --pearl-k2)