bigjme
|
|
March 06, 2014, 07:13:47 PM |
|
hmm, it seems the forum hasn't been updating me when new replies are put on, i've missed a lot
|
Owner of: cudamining.co.uk
|
|
|
ManIkWeet
|
|
March 06, 2014, 07:21:07 PM |
|
I see, and what cudaminer version are you running?
The latest one Christian posted, x64 for scrypt-jane (seems a bit faster than x86 for me). hmm, it seems the forum hasn't been updating me when new replies are put on, i've missed a lot
Ouch! I have this topic bookmarked in a way that always shows me the first unread post: https://bitcointalk.org/index.php?topic=167229.new#new
|
BTC donations: 18fw6ZjYkN7xNxfVWbsRmBvD6jBAChRQVn (thanks!)
|
|
|
bigjme
|
|
March 06, 2014, 07:24:21 PM |
|
keeps doing it for some reason. ooo well quick read through and im up to date.
on another note, im converting the amd miners to the nvidia side ;-) gtx 750 has opened up some real door ways.
even to the point of having a single pci-e go to 8 gpu's using a riser bank and as stated a few pages back, 270khash/s on pci-express 1 at x16 speed
a pci-express 3 x16 should be able to run stupid amounts of gpus on proper riser boards, aslong as the risers are powered
|
Owner of: cudamining.co.uk
|
|
|
ManIkWeet
|
|
March 06, 2014, 07:33:35 PM |
|
keeps doing it for some reason. ooo well quick read through and im up to date.
on another note, im converting the amd miners to the nvidia side ;-) gtx 750 has opened up some real door ways.
even to the point of having a single pci-e go to 8 gpu's using a riser bank and as stated a few pages back, 270khash/s on pci-express 1 at x16 speed
a pci-express 3 x16 should be able to run stupid amounts of gpus on proper riser boards, aslong as the risers are powered
Yes, as soon as the bandwidth usage by cudaminer is reduced this will work very well
|
BTC donations: 18fw6ZjYkN7xNxfVWbsRmBvD6jBAChRQVn (thanks!)
|
|
|
bigjme
|
|
March 06, 2014, 07:41:06 PM |
|
Yep.
One issue people gave was the need for such long riser cables. The one solution I gave was that usb3 will run up to 650MB/s
Pci-Express x16 runs at about 250MB/s So a usb3 cable should be able to run more then 2 cards to a main riser block. And if anyone could figure out how to get a thunderbolt cable running the link, it will run 1.25GB/s which should be fine to run 5 gpus from a single block at the same speed as paid express x16
When the bandwidth issue is fixed, in theory you should be able to run lots of cards from a single slot. Maybe an external gpus box with its own psu? Hmm be a good idea
|
Owner of: cudamining.co.uk
|
|
|
AliMan
|
|
March 06, 2014, 08:11:42 PM Last edit: March 06, 2014, 08:53:48 PM by AliMan |
|
I see, and what cudaminer version are you running?
The latest one Christian posted, x64 for scrypt-jane (seems a bit faster than x86 for me). hmm, it seems the forum hasn't been updating me when new replies are put on, i've missed a lot
Ouch! I have this topic bookmarked in a way that always shows me the first unread post: https://bitcointalk.org/index.php?topic=167229.new#newWhat was your bat when you autotuned it?
|
|
|
|
bigjme
|
|
March 06, 2014, 08:33:43 PM |
|
Who's bat?
|
Owner of: cudamining.co.uk
|
|
|
ivanlabrie
|
|
March 06, 2014, 08:41:07 PM |
|
keeps doing it for some reason. ooo well quick read through and im up to date.
on another note, im converting the amd miners to the nvidia side ;-) gtx 750 has opened up some real door ways.
even to the point of having a single pci-e go to 8 gpu's using a riser bank and as stated a few pages back, 270khash/s on pci-express 1 at x16 speed
a pci-express 3 x16 should be able to run stupid amounts of gpus on proper riser boards, aslong as the risers are powered
WHAT??? Clarify that part with the 8 gpus per riser bank plz...
|
|
|
|
bigjme
|
|
March 06, 2014, 08:44:20 PM |
|
Their making a riser bank. It's a single pci-express that slots onto the motherboard. It then has 8 riser cables coming off it to gpu sockets
Similar to the old pci-x server risers, but separated by riser cables
|
Owner of: cudamining.co.uk
|
|
|
ivanlabrie
|
|
March 06, 2014, 09:14:30 PM |
|
Their making a riser bank. It's a single pci-express that slots onto the motherboard. It then has 8 riser cables coming off it to gpu sockets
Similar to the old pci-x server risers, but separated by riser cables
Who is and how much for it? I want one :p
|
|
|
|
bigjme
|
|
March 06, 2014, 09:17:14 PM |
|
the guy setting up GPUcoin apparently has a supplier selling them, wont say who or how much.
my idea is you have a case that stores like 8 gpu's sit it next to your pc
ribbon cable from pc to gpu case. so in tern you have a gpu box and your pc. plug the gpu riser in to the pc and the gpu's would be usable on the pc :-)
|
Owner of: cudamining.co.uk
|
|
|
ManIkWeet
|
|
March 06, 2014, 09:21:42 PM |
|
What was your bat when you autotuned it?
The exact same, without -l t128x2
|
BTC donations: 18fw6ZjYkN7xNxfVWbsRmBvD6jBAChRQVn (thanks!)
|
|
|
ivanlabrie
|
|
March 06, 2014, 09:30:18 PM |
|
the guy setting up GPUcoin apparently has a supplier selling them, wont say who or how much.
my idea is you have a case that stores like 8 gpu's sit it next to your pc
ribbon cable from pc to gpu case. so in tern you have a gpu box and your pc. plug the gpu riser in to the pc and the gpu's would be usable on the pc :-)
Stop teasing, already super excited lol
|
|
|
|
bigjme
|
|
March 06, 2014, 09:31:29 PM |
|
:p your not the only one, i will buy one just for an excuse to by a new pc case ;-) i already have 4 E-ATX cases haha
|
Owner of: cudamining.co.uk
|
|
|
cbuchner1 (OP)
|
|
March 06, 2014, 09:43:06 PM |
|
committing an experimental --algo=blake feature, please run with with same kernel configurations as for keccak.
Only 56 MHash/s on GTX 750 Ti, but from here on out things can only improve... At least CPU and GPU agree on the results.
I am not sure about the difficulty to target mapping. I am currently using the same code as for Keccak. I will have to inspect the blake cpuminer to see how they do it.
what hash rates are AMD miners getting for blake256?
Christian
|
|
|
|
roy7
|
|
March 06, 2014, 09:46:34 PM |
|
Will the CUDA 6 release have any impact on mining?
|
|
|
|
ivanlabrie
|
|
March 06, 2014, 09:46:48 PM |
|
committing an experimental --algo=blake feature, please run with with same kernel configurations as for keccak.
Only 56 MHash/s on GTX 750 Ti, but from here on out things can only improve... At least CPU and GPU agree on the results.
I am not sure about the difficulty to target mapping. I am currently using the same code as for Keccak. I will have to inspect the blake cpuminer to see how they do it.
Christian
Excellent news! man never rests...major props Christian.
|
|
|
|
cbuchner1 (OP)
|
|
March 06, 2014, 09:48:24 PM |
|
Will the CUDA 6 release have any impact on mining?
so far it's a negative impact. So CUDA 6 RC is a no-go. There are serious performance regressions for some kernels, and no additional gain for code compiled specifically for the 750 Ti (Maxwell). Christian
|
|
|
|
djm34
Legendary
Offline
Activity: 1400
Merit: 1050
|
|
March 06, 2014, 10:05:51 PM |
|
committing an experimental --algo=blake feature, please run with with same kernel configurations as for keccak.
Only 56 MHash/s on GTX 750 Ti, but from here on out things can only improve... At least CPU and GPU agree on the results.
I am not sure about the difficulty to target mapping. I am currently using the same code as for Keccak. I will have to inspect the blake cpuminer to see how they do it.
what hash rates are AMD miners getting for blake256?
Christian
I just tried on one pool Blakecoin EU1 pool, I get only boo's. the hashrate would be max 130Mhash, I just checked with the big amd electron's sucker (which doesn't seem to use that much on it), ouch 3.25GHash/s
|
djm34 facebook pageBTC: 1NENYmxwZGHsKFmyjTc5WferTn5VTFb7Ze Pledge for neoscrypt ccminer to that address: 16UoC4DmTz2pvhFvcfTQrzkPTrXkWijzXw
|
|
|
cbuchner1 (OP)
|
|
March 06, 2014, 10:13:28 PM Last edit: March 06, 2014, 10:24:29 PM by cbuchner1 |
|
I just tried on one pool Blakecoin EU1 pool, I get only boo's. the hashrate would be max 130Mhash, I just checked with the big amd electron's sucker (which doesn't seem to use that much on it), ouch 3.25GHash/s
this is the code from the blakecoin CPU miner... so the line on the bottom applies to Blake256. I fixed this in cudaminer now. if (opt_algo == ALGO_SCRYPT) diff_to_target(work->target, sctx->job.diff / 65536.0); else diff_to_target(work->target, sctx->job.diff);
About the hashing speed: This is a straight port of the CPU code running mostly unoptimized in CUDA. Bad speed is kind of expected, although 50 MHash/s is a start... If hashes are still getting rejected now, there must be another reason (maybe endianness problems....) Christian
|
|
|
|
|