-->

Thursday, November 10, 2022

Rattly, the sequel

 Today Nov. 10, Windows update installed a plethora of updates, without affecting Timespy noticeably.




 I'd very much like to point out how crowded-together the updates are this week, and pointedly remark that the first update (on monday?) messed up Timespy, and the Current one left it alone (aka "fixed" whatever was causing a problem)

Right this exact second, having updated my 22621 to 22623, the Timespy program has decided to slow down again.
How granular are my system-restores (Can I return to an hour ago and still maintain most of the updates?)
I'd like to insert a line here (somewhere) about "madness," but this is still being edited, and it might be premature.

22621 also runs slow! Timespy is royally pissed off but will not say why.

*You* might loudly toot that it is a driver problem, and in fact the driver both from microsoft and the one from AMD are very recent, however I've tried both with the same results.
Should I try a much older driver? You'd think so, I'll see.
No help.
And the really mysterious part is the part where it appears to run normally before it reverts to slowness.
There are security and "stack" updates I'd miss if I went ahead and reinstalled my old ISO file.
I'll wait.
No, Actually, I won't, I'll reinstall the (only) ISO I've got, and See if it brings back the speed, using the latest AMD driver.
TBA but if this works, I might stop or lower my beta-status.
No one reads this stuff anyway.
I only hope someone else, who rates, coincidentally discovers this before the final version is released, 
Or,
 3Dmark sees what's happening and issues a patch (of some sort.)
Either way, I'm speaking before I have finished my installation, so like I said, "TBA".
F'n-A, the speed is OK!!
No one cares, 
but hey, that's Okay.
*I*m happy.

Back to mediocrity, unless I update.
Will I need to *test* every single update from now on?
This week's updates tanked the speed, but next week, who knows.


A really smart troubleshooter would discover the exact file or registry-setting doing this.
Please note, My complaint is, 3dmark-Timespy runs at half-speed, not just a little slower.
Maybe that's significant (two programs running at once?) maybe not.
Oct 2021, CPU's with two CCD's and SMT-enabled, ran worse, hmm, but since then supposedly they updated stuff (I thought that's what we were talking about, "Updates".)
Let us imagine (wildly speculate) that Updates to CCD's are so obscure that updates from microsoft might not have them in every version.
But (dammit) I'm lazy, and would not want to DISable one of the CCD's and DISable "SMT", but I will, next time this happens.
The undead SMT-etc syndrome as it relates to 7950:

The 7950 has two CCD's (I think) and I know my 3950 does.
Whaddifff....
I disable a ccd (But my system is sitting pretty now on the old version, why change it?)
But it's a possible solution. Or another myth.





Apparently changing SMT-Control to "disabled" is a biggie deal on my system, the BIOS knows to shut-down power and restart. ("Shut-down-power", are you even listening anymore??) So OK tests need to be run!! I've got no more virtual cores!! "Simultaneous multithreading" is off, says Ryzen Master! To what end?? 
Well, *yeah*, um, it's technically faster, uh, ?? The CPU test is short and very boring.
The GPU says "meh"
And its biggie-brother, "extreme" (@3840*2160)

"Kiss my nose and Lick my toes, Worship me!"
Well ok, nothing like that, but I'll subsist on the meager praise.

That'll do, pig, that'll do




Anyway, forums have people who look a little too by-the-book insisting you do NOT disable SMT. (Ya wanna know WHY????) 
Hackers tend to give the exact WRONG advice just to make stuff easier to hack... if an dishonest hardworking hacker is probing ports in your neighborhood, (for example) they want to see "open" or "closed", not get no answer, it (according to one hacker) *inflames* them and they'll choose your computer to attack.
So, when they tell me not to disable SMT, because (ramalama-folderol) I become suspicious.
Consider a lowly(?) CPU with 8 cores, 8 threads.
I've got 16 *cores* and do not actually need the accompanying 16 threads...
so...um,...I made my point, even if I don't know how to state it.
3AM with a very sore butt from sitting too long, I tend to digress...
So I'll summarize...Updates=slower, sometimes. But Why? No one knows (no one googleable)




Longstanding tradition (since 1985) to hack spy on neighbors and "evil" companies.
(Thanks, Angelina)


But I would not be stewpid, I try at least to make my PC and my apartment *so Boring* you'll pass it by.
Sniffing around? Go away, nothing to smell here
Peepers, long standing tradition upholders, take (might take) umbrage I'd even mention them.
Bah!
I was brought up to MYOB, ignore the guy crossing the street with his c*ck hanging out peeing, ignore the subway ranter....
(Random trendy-type)


Anyway SMT apparently isn't something I'd need, and not being a server, something no one would want. 
(That's another question, why are people all hot to run servers?)
Pay for protection











No comments: