Core 1.48 and SSE - BUG

[Rav]

I think I've found a bug with the new core. When it came out, I deleted my old cores and restarted the clients with -advmethods and -forceasm as usual. I saw Extra SSE boost OK and was happy. Well I checked my logs this morning and found something a bit upsetting. After my rigs finished their current WU's, the next one they picked up only used 3DNow! boost! If I shut down and restart the client it starts that unit with SSE again. But when it finishes and gets a new unit it goes back to 3DNow. I hope I don't have to restart my clients every WU

Anyone else see this happen?

-Rav

 

[RoadWarrior]

That's not a bug it's a feature. Honest. See the thread below about the new core.

 

[Rav]

I see the Beta 3.25 client fixes this bug. From the changelog:

- If -forceasm is used, then on every execution of the core until the
client is restarted, SSE optimizations will be used if possible (if
the chip is an AMD chip, then be default, without the flag, it would
not be chosen over 3DNow). Prior to this client, this effect would have
held true only for the first work unit run after starting the client.

-Rav

 

[Xevuhtess7]

is this bug only with athlonxp's? the sig in my rig (p4) shows the extra sse boost ok thing... cept im 95/100 frames done so when i ge thome ill finish those 5 frames and see if it doens't show up next wu..... also is the new core the only one that should say "extra sse boost" thingy? cause my client might have downloaded the new core automatically.... and i thought the clients weren't set to download the new core yet? (yesterday i saw my client redownload the core for some reason upon receiving the new wu...)

 

[nikhsub1]

is this bug only with athlonxp's? the sig in my rig (p4) shows the extra sse boost ok thing... cept im 95/100 frames done so when i ge thome ill finish those 5 frames and see if it doens't show up next wu..... also is the new core the only one that should say "extra sse boost" thingy? cause my client might have downloaded the new core automatically.... and i thought the clients weren't set to download the new core yet? (yesterday i saw my client redownload the core for some reason upon receiving the new wu...)

First guys, this is NOT a bug, it is by design... the new 3.25 client fixes this if desired. This does NOT affect the Pentiums as SSE optimizations are ON by default with all Pentiums that support it.

 

[Fever]

I cannot, for the life of me, get SSE working again. I've got the new 1.48 core, the new 3.25 client, and it's still no go!