<html>
<head>
<base href="https://bugs.freedesktop.org/" />
</head>
<body>
<p>
<div>
<b><a class="bz_bug_link
bz_status_NEW "
title="NEW --- - bfgminer --scrypt on 7xxx+"
href="https://bugs.freedesktop.org/show_bug.cgi?id=72785#c3">Comment # 3</a>
on <a class="bz_bug_link
bz_status_NEW "
title="NEW --- - bfgminer --scrypt on 7xxx+"
href="https://bugs.freedesktop.org/show_bug.cgi?id=72785">bug 72785</a>
from <span class="vcard"><a class="email" href="mailto:tstellar@gmail.com" title="Tom Stellard <tstellar@gmail.com>"> <span class="fn">Tom Stellard</span></a>
</span></b>
<pre>(In reply to <a href="show_bug.cgi?id=72785#c0">comment #0</a>)
<span class="quote">> bfgminer --benchmark without scrypt seems to be working fine now.
>
> But I think many people would rather compute scrypt right now on their 7xxx+
> GPUs, just considering the recent rumors that the radeon 7950 is getting
> sold out because of litecoin miners.
>
> With mesa/llvm it doesn't work yet.
>
> llvm 3.5 197392, mesa and stuff are very recent git builds.
>
> The kernel is
> <a href="https://github.com/luke-jr/bfgminer/blob/bfgminer/scrypt130511.cl">https://github.com/luke-jr/bfgminer/blob/bfgminer/scrypt130511.cl</a>
>
> bfgminer is started with
>
> bfgminer -v1 --scrypt -S opencl:auto --url=<pool>
> --userpass=<user:worker:pass>
>
> The error is:
>
> Initialising kernel scrypt130511.cl without bitalign, 1 vectors and worksize
> 256
>
> [...]
>
> LLVM ERROR: Cannot select: 0x7f0b7cbd40a0: ch = br_cc 0x7f0b7cd32bb0,
> 0x7f0b7cc28c20, 0x7f0b7cd32fb0, 0x7f0b7cc2bc30, 0x7f0b7cc2ad30 [ORD=19251]
> [ID=4253]
> 0x7f0b7cd32fb0: i1,ch = llvm.SI.loop 0x7f0b7cc2c030:1, 0x7f0b7cd333b0,
> 0x7f0b7cc2c030 [ORD=19250] [ID=4250]
> 0x7f0b7cd333b0: i64 = TargetConstant<3431> [ID=132]
> 0x7f0b7cc2c030: i64,ch = llvm.SI.if.break 0x7f0b7cf2f3e0,
> 0x7f0b7cc2b530, 0x7f0b7cc2c530, 0x7f0b7cd339b0 [ORD=19249] [ID=4249]
> 0x7f0b7cc2b530: i64 = TargetConstant<3428> [ID=129]
> 0x7f0b7cc2c530: i1 = setcc 0x7f0b7cc7dd60, 0x7f0b7cc80760,
> 0x7f0b7cd337b0 [ORD=19248] [ID=212]
> 0x7f0b7cc7dd60: i64 = add 0x7f0b7cc81560, 0x7f0b7cc81360 [ORD=13153]
> [ID=187]
> 0x7f0b7cc81560: i64,ch = CopyFromReg 0x7f0b7c0535d8,
> 0x7f0b7cc81660 [ORD=13147] [ID=149]
> 0x7f0b7cc81660: i64 = Register %vreg53 [ID=1]
> 0x7f0b7cc81360: i64 = Constant<1> [ID=2]
> 0x7f0b7cc80760: i64 = Constant<4> [ID=4]
> 0x7f0b7cd339b0: i64,ch = CopyFromReg 0x7f0b7c0535d8, 0x7f0b7cd342b0
> [ORD=19249] [ID=185]
> 0x7f0b7cd342b0: i64 = Register %vreg52 [ID=130]
> 0x7f0b7cc2bc30: i1 = Constant<-1> [ID=133]
> In function: search</span >
This patch:
<a href="http://cgit.freedesktop.org/~tstellar/llvm/commit/?h=master-testing-si&id=41d1f890bc2095a1354c4502cc97e30b3288ba1a">http://cgit.freedesktop.org/~tstellar/llvm/commit/?h=master-testing-si&id=41d1f890bc2095a1354c4502cc97e30b3288ba1a</a>
should fix the crash in an unmodified kernel. However, you will probably still
see the error from <a href="show_bug.cgi?id=72785#c3">comment 3</a>.
I need a test case in order to commit this patch. Could you post the output of
R600_DEBUG=cs from an unmodified kernel *without* this patch applied.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are the assignee for the bug.</li>
</ul>
</body>
</html>