Post Your GPU Benchmarks for WPA

[Ntrain20k]

Anyone got the AMD RX 7900 XT and XTX WPA speeds?

Also updated my GPU speed comparison text above with this. Worth reading!

"Another note is the GPUs run a lot cooler when reducing the POWER (obviously). Much better way to run jobs I think. More efficient and cooler. I never run at 100% POWER any more. Doing this also extends the lifespan of the GPUs, as less thermal damage due to over-heating."

I have an XTX on the way.

 

[thanost]

None of my posts have to be approved. Looking at your post history I can see why they might have you on approval required.....

I agree
It is the price you must pay for honest information about hash/Wattt

 

[Ntrain20k]

Benchmark: RX 7900 XTX

OpenCL API (OpenCL 2.1 AMD-APP (3516.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #1: AMD Radeon(TM) Graphics, 6080/12284 MB (5886 MB allocatable), 1MCU
* Device #2: AMD Radeon RX 7900 XTX, 24448/24560 MB (24560 MB allocatable), 48MCU

Benchmark relevant options:
===========================
* --workload-profile=4

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 29431 H/s (277.14ms) @ Accel:2048 Loops:256 Thr:64 Vec:1
Speed.#2.........: 1548.5 kH/s (124.06ms) @ Accel:64 Loops:1024 Thr:256 Vec:1
Speed.#*.........: 1578.0 kH/s

Started: Sat Feb 18 20:13:53 2023
Stopped: Sat Feb 18 20:14:13 2023

2628 MHz 374 watts

Powercolor red devil, stock settings.



OpenCL API (OpenCL 2.1 AMD-APP (3516.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #1: gfx1036, 12220/12284 MB (11772 MB allocatable), 1MCU
* Device #2: gfx1100, 24496/24560 MB (24560 MB allocatable), 48MCU

Benchmark relevant options:
===========================
* --workload-profile=4

Hashmode: 22000 - WPA-PBKDF2-PMKID+EAPOL (Iterations: 4095)

Speed.#1.........: 29573 H/s (276.53ms) @ Accel:1024 Loops:512 Thr:64 Vec:1
Speed.#2.........: 1581.6 kH/s (247.72ms) @ Accel:1024 Loops:512 Thr:64 Vec:1
Speed.#*.........: 1611.2 kH/s

Started: Sat Feb 18 20:23:29 2023
Stopped: Sat Feb 18 20:23:36 2023

351 watts with some undervolting.

 

[blandyuk]

Awesome! About time we had a WPA bench with the RX 7900 XTX. Updated my Excel sheet below. Also added RTX 4090 twice, with both wattages as unsure:



If the RTX 4090 does use 600 watts with hashcat, then the RX 7900 XTX is up there from an efficiency point-of-view.

 

[Hashpup2222]

Can anyone point in the right direction on these builds there looking around 15 thousand usd builds I would assume would like to replicate these top builds, I’m not experienced but lambalabs towers are like 80 thousand so if anyone can help me that would be nice

 

[blandyuk]

Presume you mean a cracking rig. Cheaper to build your own and use an open air setup like mine. Search ebay for: "ethereum mining frame"
(Use rear fans at the back to help cool the cards, makes a big difference).

Then you need motherboard: https://www.ebay.co.uk/itm/234640348251

Good CPU and RAM. SSD HDD, PCIe risers, mining PSU(s) and then it's just the GPU cards, easy! If you buy older cards, u can get them cheap. Can pick up RTX 2080 Ti cards for £350-£400 so ~$480 each.

You'll get a 10x RTX 2080 Ti rig for ~£4000, maybe slightly more. Get 32GB of RAM as you'll need it with more cards. hashcat will allocate a lot with big setups depending on attack type / parameters.

Make sure you buy GOOD GPU cards. Use Gamer Nexus reviews to decide.

https://www.youtube.com/channel/UChIs72whgZI9w6d6FhwGGHA

Additional:

Always consider the power usage for cracking rigs. Get PCIe risers with 6-pin PCIe power connectors. The GPUs will likely use 2x 8-pin, so that's 3 PCIe power connecter per card! My two 1300watt PSUs have 12x 6+2-pin connectors, so will power 4x RTX 2080 Ti @ 250 watts each, (8x for the cards, 4x for the risers). If you have a 10 card rig, that's 3x PSUs.

Most GPU rig issues are power related. You need to get power as even as possible. Also, don't get multi-rail PSUs, get single-rail if you can. This relates to the 12v rail in the PSU. Multi-rail PSUs tend to overload easily as power cannot be distributed as easy. Single 12v rail PSUs don't have this issue.

 

[PapaSmurf]

CUDA API (CUDA 12.0)
====================
* Device #1: NVIDIA GeForce RTX 4070 Ti, 11052/12281 MB, 60MCU
* Device #2: NVIDIA GeForce RTX 3080 Ti, 11086/12287 MB, 80MCU

OpenCL API (OpenCL 3.0 CUDA 12.0.147) - Platform #1 [NVIDIA Corporation]
========================================================================
* Device #3: NVIDIA GeForce RTX 4070 Ti, skipped
* Device #4: NVIDIA GeForce RTX 3080 Ti, skipped

Benchmark relevant options:
===========================
* --workload-profile=3

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........:  1222.0 kH/s (49.36ms) @ Accel:8 Loops:1024 Thr:512 Vec:1
Speed.#2.........:  1049.5 kH/s (76.66ms) @ Accel:8 Loops:1024 Thr:512 Vec:1
Speed.#*.........:  2271.5 kH/s

4070 Ti is at 245W
3080 Ti is at 355W
according to GPU-Z
Both at stock clocks

 

[blandyuk]

Nice bench for the RTX 4070 Ti, I get an efficiency of (1222000 / 245) = 4987.8, which is very good.

 

[zen101]

OpenCL API (OpenCL 2.1 AMD-APP (3570.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #1: AMD Radeon RX 7900 XTX, 24448/24560 MB (20876 MB allocatable), 48MCU
* Device #2: AMD Radeon RX 6900 XT, 16256/16368 MB (13912 MB allocatable), 40MCU

Benchmark relevant options:
===========================
* --backend-devices-virtual=1
* --workload-profile=4

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 1482.8 kH/s (130.57ms) @ Accel:512 Loops:1024 Thr:32 Vec:1
Speed.#2.........: 1110.4 kH/s (291.81ms) @ Accel:256 Loops:1024 Thr:128 Vec:1
Speed.#*.........: 2593.2 kH/s

Started: Sun Jul 16 15:11:46 2023
Stopped: Sun Jul 16 15:12:13 2023

 

[zen101]

OpenCL API (OpenCL 2.1 AMD-APP (3570.0)) - Platform #1 [Advanced Micro Devices, Inc.]
=====================================================================================
* Device #1: AMD Radeon RX 6700 XT, 12160/12272 MB (10213 MB allocatable), 20MCU
* Device #2: AMD Radeon RX 6700 XT, 12160/12272 MB (10213 MB allocatable), 20MCU

Benchmark relevant options:
===========================
* --backend-devices-virtual=1
* --workload-profile=4

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 643.7 kH/s (251.33ms) @ Accel:512 Loops:512 Thr:128 Vec:1
Speed.#2.........: 647.8 kH/s (250.55ms) @ Accel:512 Loops:512 Thr:128 Vec:1
Speed.#*.........: 1291.5 kH/s

Started: Sun Jul 16 15:45:44 2023
Stopped: Sun Jul 16 15:46:11 2023

 

[Atlantis555]

CUDA API (CUDA 12.2)
====================
* Device #1: NVIDIA GeForce RTX 2060 SUPER, 7165/8191 MB, 34MCU

Benchmark relevant options:
===========================
* --optimized-kernel-enable
* --workload-profile=3

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 449.4 kH/s (75.98ms) @ Accel:16 Loops:512 Thr:512 Vec:1

 

[control]

CUDA API (CUDA 12.1)
====================
* Device #1: NVIDIA GeForce RTX 4090, 23008/24563 MB, 128MCU
* Device #2: NVIDIA GeForce RTX 4090, 23008/24563 MB, 128MCU
* Device #3: NVIDIA GeForce RTX 2080 Ti, 10121/11263 MB, 68MCU

OpenCL API (OpenCL 3.0 CUDA 12.1.112) - Platform #1 [NVIDIA Corporation]
========================================================================
* Device #4: NVIDIA GeForce RTX 4090, skipped
* Device #5: NVIDIA GeForce RTX 4090, skipped
* Device #6: NVIDIA GeForce RTX 2080 Ti, skipped

OpenCL API (OpenCL 3.0 ) - Platform #2 [Intel(R) Corporation]
=============================================================
* Device #7: Intel(R) UHD Graphics 770, 26112/52311 MB (2047 MB allocatable), 32MCU

OpenCL API (OpenCL 2.1 WINDOWS) - Platform #3 [Intel(R) Corporation]
====================================================================
* Device #8: 13th Gen Intel(R) Core(TM) i9-13900KS, skipped

Benchmark relevant options:
===========================
* --optimized-kernel-enable

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 2606.1 kH/s (49.83ms) @ Accel:16 Loops:512 Thr:512 Vec:1
Speed.#2.........: 2148.7 kH/s (54.32ms) @ Accel:32 Loops:512 Thr:256 Vec:1
Speed.#3.........: 888.1 kH/s (76.79ms) @ Accel:32 Loops:512 Thr:256 Vec:1
Speed.#7.........: 23973 H/s (84.18ms) @ Accel:256 Loops:64 Thr:16 Vec:1
Speed.#*.........: 5666.9 kH/s

Started: Fri Jul 28 22:58:15 2023
Stopped: Fri Jul 28 22:58:49 2023

 

[174region174]

CUDA API (CUDA 12.0)
====================
* Device #1: NVIDIA GeForce RTX 3080 Ti, 11086/12287 MB, 80MCU
* Device #2: NVIDIA GeForce GTX 1080 Ti, 10216/11263 MB, 28MCU
* Device #3: NVIDIA GeForce GTX 1080 Ti, 10217/11263 MB, 28MCU
* Device #4: NVIDIA GeForce GTX 1080 Ti, 10217/11263 MB, 28MCU
* Device #5: NVIDIA GeForce GTX 1080 Ti, 10217/11263 MB, 28MCU

OpenCL API (OpenCL 3.0 CUDA 12.0.89) - Platform #1 [NVIDIA Corporation]
=======================================================================
* Device #6: NVIDIA GeForce RTX 3080 Ti, skipped
* Device #7: NVIDIA GeForce GTX 1080 Ti, skipped
* Device #8: NVIDIA GeForce GTX 1080 Ti, skipped
* Device #9: NVIDIA GeForce GTX 1080 Ti, skipped
* Device #10: NVIDIA GeForce GTX 1080 Ti, skipped

Benchmark relevant options:
===========================
* --workload-profile=3

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 1132.7 kH/s (71.78ms) @ Accel:16 Loops:1024 Thr:256 Vec:1
Speed.#2.........: 669.9 kH/s (84.25ms) @ Accel:16 Loops:1024 Thr:512 Vec:1
Speed.#3.........: 661.7 kH/s (85.33ms) @ Accel:16 Loops:1024 Thr:512 Vec:1
Speed.#4.........: 626.9 kH/s (89.00ms) @ Accel:16 Loops:1024 Thr:512 Vec:1
Speed.#5.........: 651.9 kH/s (86.61ms) @ Accel:16 Loops:1024 Thr:512 Vec:1
Speed.#*.........: 3743.2 kH/s

Started: Sun Oct 08 16:34:47 2023
Stopped: Sun Oct 08 16:35:30 2023

 

[174region174]

CUDA API (CUDA 12.0)
====================
* Device #1: NVIDIA GeForce RTX 3080 Ti, 11788/12045 MB, 80MCU
* Device #2: NVIDIA GeForce RTX 3080 Ti, 11788/12045 MB, 80MCU
* Device #3: NVIDIA GeForce GTX 1080 Ti, 10827/11169 MB, 28MCU
* Device #4: NVIDIA GeForce GTX 1080 Ti, 11027/11172 MB, 28MCU
* Device #5: NVIDIA GeForce GTX 1080 Ti, 11027/11172 MB, 28MCU

OpenCL API (OpenCL 3.0 CUDA 12.0.151) - Platform #1 [NVIDIA Corporation]
========================================================================
* Device #6: NVIDIA GeForce RTX 3080 Ti, skipped
* Device #7: NVIDIA GeForce RTX 3080 Ti, skipped
* Device #8: NVIDIA GeForce GTX 1080 Ti, skipped
* Device #9: NVIDIA GeForce GTX 1080 Ti, skipped
* Device #10: NVIDIA GeForce GTX 1080 Ti, skipped

OpenCL API (OpenCL 3.0 PoCL 3.1+debian Linux, None+Asserts, RELOC, SPIR, LLVM 15.0.6, SLEEF, DISTRO, POCL_DEBUG) - Platform #2 [The pocl project]
==================================================================================================================================================
* Device #11: pthread-haswell-Intel(R) Core(TM) i7-5775C CPU @ 3.30GHz, skipped

Benchmark relevant options:
===========================
* --workload-profile=3

-------------------------------------------------------------
* Hash-Mode 22000 (WPA-PBKDF2-PMKID+EAPOL) [Iterations: 4095]
-------------------------------------------------------------

Speed.#1.........: 1122.3 kH/s (72.16ms) @ Accel:16 Loops:1024 Thr:256 Vec:1
Speed.#2.........: 1122.2 kH/s (72.17ms) @ Accel:16 Loops:1024 Thr:256 Vec:1
Speed.#3.........: 647.2 kH/s (85.74ms) @ Accel:32 Loops:512 Thr:512 Vec:1
Speed.#4.........: 644.7 kH/s (85.98ms) @ Accel:32 Loops:512 Thr:512 Vec:1
Speed.#5.........: 645.0 kH/s (87.41ms) @ Accel:16 Loops:1024 Thr:512 Vec:1
Speed.#*.........: 4181.4 kH/s

Started: Mon Oct 9 08:25:42 2023
Stopped: Mon Oct 9 08:25:51 2023

 

[Hashpup2222]

wow those last three post are insane the most ill be able to do is 2880 kh/s