[Dev] NeoScrypt GPU Miner - Public Beta Test
-
Was that your first time using my fixed kernel on 14.9?
Actually yes… i guess i am being a bit slow on those updates :D Good job! Thanks for your involvement
-
Actually yes… i guess i am being a bit slow on those updates :D Good job! Thanks for your involvement
No problem; you should be getting more hash soon!
-
No problem; you should be getting more hash soon!
Let me step right up and personally thank you for the development you have done on this.
Post a btc address and I’ll send you a couple satoshi, or post a guncoin address and I’ll send you a couple thousand. ;)
-
It’s 19.5KH/s now on a HD6970. FastKDF and BLAKE2s have been cleaned up and optimised, memory requirements reduced.
Yeah, I’ve mentioned this in my white paper. Not sure if it’s of any use for mining.
It is, but that’s not what concerns me now. With FastKDF removed, the kernel gets reduced in size by ~60% and outputs 30KH/s.That’s a big overhead, but not critical and I’ve expected more out of ChaCha + Salsa. With ChaCha only enabled, it’s 58KH/s and with Salsa only = 56KH/s. Scalar Salsa isn’t supposed to be about as fast as vectorised ChaCha. It’s clearly scalar because the AMD compiler isn’t really smart and the kernel size is about double of ChaCha only size. Anyway, there is a huge bottleneck somewhere and it needs to be identified.
Don’t work for me now. Cards in rig: 6950,6870,5870, miner 3.7.7b. Screen:
-
Don’t work for me now. Cards in rig: 6950,6870,5870, miner 3.7.7b. Screen:
Post more information.
For example:
Windows version and if it is 64 or 32-bit
AMD Catalyst drivers
what worksize are you using?
have you set the following environmental vars?
GPU_MAX_ALLOC_PERCENT=100
GPU_USE_SYNC_OBJECTS=1
also, upgrade to 3.7.7c or use sgminer
-
I submitted a bug report to sgminer-dev github in regards to the apparent worksize issue in sgminer-dev binaries on Win64.
-
Post more information.
For example:
Windows version and if it is 64 or 32-bit
AMD Catalyst drivers
what worksize are you using?
have you set the following environmental vars?
GPU_MAX_ALLOC_PERCENT=100
GPU_USE_SYNC_OBJECTS=1
also, upgrade to 3.7.7c or use sgminer
Yep, i forgot this settings: GPU_MAX_ALLOC_PERCENT=100 , GPU_USE_SYNC_OBJECTS=1… on sgminer5 now works fine, thanks! 24 kh/s on radeon 6870 :))
-
Let me step right up and personally thank you for the development you have done on this.
Post a btc address and I’ll send you a couple satoshi, or post a guncoin address and I’ll send you a couple thousand. ;)
I will once the release happens; still waiting on a withdrawal.
-
Don’t work for me now. Cards in rig: 6950,6870,5870, miner 3.7.7b. Screen:
MAX_GLOBAL_THREADS equals to CONCURRENT_THREADS in SGminer and other non-customised miners. This is --thread-concurrency.
-
Great work , neoscrypt_vliw.cl v2 is very good ,my 6770 card can work. :)
-
I updated my drivers from 14.4 to 14.9 on an HD 6950 with shaders unlocked (Windows 7 64 bit). My hashrates increased a little, but I stopped getting accepteds. I tried Wolf0s’ sgminer build, and cgminer 3.7.8. I even tried redownloading sgminer to start with a clean slate.
I am using GPU_MAX_ALLOC_PERCENT=100
GPU_USE_SYNC_OBJECTS=1
I tried reducing intensity, lowering -w, and some other stuff.
I uninstalled the 14.9 and reinstalled 13.12 drivers (I didn’t really love 14.4) and I get accepteds again.
Still at 16 Kh/s when newer cards get almost 10X that. On both scrypt and groestl, HD 6950s are good for about 1/2 what a 290X gets (450 Kh/s scrypt and 7.5 Mh/s groestl.) But it is much better than it was, so thanks Wolf0 and Ghostlander. Hopefully we’ll get up to 50-75 Kh/s soon on the older hardware.
-
Hi, so im using CGMiner 3.7.8 with 14.9 7970 with R9 280X bios: cgminer --neoscrypt -I 15 -g 2 -w 48 --gpu-memclock 1400 --gpu-engine 1000
Im getting 110Kh/s, Do you guys think this is about right? Or should i be getting more?
Cheers
-
Thats about right, depends on the memory type, some of my 7970s/280x wont go above 135 at 1125/1500 while other ramp all the way up to 160-170 khs
-
There is no point of using HD5000 and HD6000 series with v14.x drivers. They are all optimised for the GCN cards. I’m fine with v13.1 and SDK v2.6. v13.4 also seems good. v13.8beta was crap. Didn’t even bother with v13.12 and newer.
-
There is no point of using HD5000 and HD6000 series with v14.x drivers. They are all optimised for the GCN cards. I’m fine with v13.1 and SDK v2.6. v13.4 also seems good. v13.8beta was crap. Didn’t even bother with v13.12 and newer.
I’m pretty sure I have 13.12 on my rig with the 69xx cards and maybe this is why I only saw a 1.25kh/s increase in hash with the v2 over v1?
-
There is no point of using HD5000 and HD6000 series with v14.x drivers. They are all optimised for the GCN cards. I’m fine with v13.1 and SDK v2.6. v13.4 also seems good. v13.8beta was crap. Didn’t even bother with v13.12 and newer.
What if we have a mult-card setup with both GCN and pre-GCN cards?
-
Here you guys go: https://ottrbutt.com/tmp/neoscrypt.cl
-
[2014-11-10 22:25:50] Started cgminer 3.7.8
[2014-11-10 22:25:50] Network diff set to 551K
[2014-11-10 22:25:50] Stratum from pool 0 detected new block
[2014-11-10 22:25:50] Stratum from pool 0 requested work restart
[2014-11-10 22:25:52] Error -11: Building Program (clBuildProgram)
[2014-11-10 22:25:52] :486:16: error: attribute requires 1 argument(s)
__attribute__((reqd_work_group_size(WORKSIZE, 1, 1)))
^
:512:11: warning: incompatible pointer types passing ‘ulong16 __attribute__((add
ress_space(1))) *’ to parameter of type 'uint16 __attri
[2014-11-10 22:25:52] Failed to init GPU thread 0, disabling device 0
[2014-11-10 22:25:52] Restarting the GPU from the menu will not fix this.
[2014-11-10 22:25:52] Try restarting cgminer.
Press enter to continue: -
Here you guys go: https://ottrbutt.com/tmp/neoscrypt.cl
Thank you again Wolf! My hash rates are so much higher now!
-
Here you guys go: https://ottrbutt.com/tmp/neoscrypt.cl
Wolf0, that one didn’t help on pre-GCN cards, at least on mine. Hashrate went from 16.5 Kh/s to 13 Kh/s. Not that you said it would help old cards, but for those of us with older hardware, this appeared to be a step in the wrong direction.
