|
|
A mixed bag of raging
squirrels...
Roelof
has been gnawing away at the bark on his version tree of SETISpy
and released v 2.9.2 which has a several bug fixes (minor) and one
very useful new option. Now you can enable your personal stats to be
updated when you transmit results. A simple thing but ultimately
pleasing little addition. Before more people mention it (thanks to
those that have) the benchmark results table is down at the moment
pending a re-adjustment by Rat
Bastard that
will hopefully start showing the times I occasionally mention,
that'll be nice.
A few bits and pieces to intrigue and inform or just fill up space
depending on how you look at these things. Perhaps you have come
across the SETI
stats at Roving mouse
but if not they are worth a quick look to give you an idea of the
bigger SETI crunching picture. Just when I feel a glow for
mentioning the VIA 4 in 1 drivers making it to 4.24 a while back I
discover that 4.25 is making the rounds! Typical, if you fancy the download
(800kb) and then head to viahardware
(Sept9) for a bit of info. Perhaps this will turn into a 'chase the
latest beta DirectX 8' type situation but I hope not. Want to know
who made that cheapo RAM you're saddled with (cheap as in it won't
do 150FSB at CAS2) try hunting down a manufacturer or two at chipmunk.
Also in memory vein how about a visual
representation of RAM timings and dataflow, hooked aren't you, I
can tell. It's a neat little java applet showing PIII and PenPro
signal-lines and bytebursts that repays a little watching and
figuring out. So that's what some of those tweak
bios, WPCRedit
and AGPinfo
acronyms refer to...
Otherwise it is still very quiet in this area of SETI newsdom. As
you might expect most of the interesting things have been weeded out
and become mainstream knowledge hence my weekly or two updates are
mainly fuelled by the submission page entries. These still trickle
steadily in so lets head into some times...
In with a slightly slow-ish 18:25 is zmartin
running the Cli 2.4 on Win98 with a 350 K6-2. zmartin
also submitted 11:20 for a Celeron at 333, hamstrung with a 66MHz
bus (I hope I will be hearing the screams of low multiplier
overclocking soon) there's real potential here for further delights.
A Mac clone (Power Center Pro with a 450 G3 processor)
produced a 9:14 for jonlaye
which is adequately solid (wouldn't want to commit myself to saying
good) for what is likely to be a conservatively specified machine.
Moving up a serious speed notch courtesy of Poyares
SGI machine. This houses a MIPS-4 beastie (likely to be a R12K
considering the quoted 300MHz , 64 bit RISC instruction set,
superscalar architecture, most munchy) that grinds through bench
WU's in 3:02. A ray of brightness in an otherwise dull period.
After we sorted out the time discrepancies, in an exceptionally
speedy TBird submission, Chad
asked me not to mention
anything about the matter. He seemed quite down about the whole
episode but there's no need to be, genuine mistakes are just
discontinuities on the learning curve. Far better to have misplaced
enthusiasm than 'seen it , done it, fed up with it' type of world
weariness, pass me the Amyl Nitrite! Also in the 'staggering times'
category smokvas'
Celeron is either a
werewolf in sheep's clothing or a mistake. Could you get in touch,
thanks. In similar vein Alexandre
your revised time for
your Mac G4 is still way too big, somewhere between 5 and 7 hours
would make sense. Needs some investigation at your end or send the
result.sah to me.
I rarely digress from SETI matters so for a change... I gave blood
today (literally) and received 2 packets of digestives and some tea
for my pumped donation. After injury my running & swimming are
coming back well so I figured it was time to repay my good fortune
by giving a few cells to those who need them more than me. Final
observation concerns the BBC commentary team (so called
professionals) that produced some disgraceful remarks about the
swimming of three competitors in one of the early Wednesday 50m
women's freestyle heats at the Olympics. The fact that these people
were there at all was fantastic (one had never swum 50m before (!)
having only started in January) and somehow so fitting a contrast to
all those elite athletes. Trying is as important as winning. Well I
got that off my chest. God help us if ET doesn't live up to our
smugly superior expectations.
Max out.
Driven to distraction on Tracy
island...
So it's new version time at Mike Obers'
shed, SetiDriver
is now available at 1.5 with loads of minor fixes and improvements.
Looking at the history
page shows just how much this piece of caching brilliance has
improved since my very reserved view of it on first appearance.
Continuing in the Spy/Driver vein, call me jaded and world weary but
occasionally I do get a little excited and the most recent trouser
tremble came from a snippet on alt.sci.seti. In answer to the
comment "Hmmm. Maybe Mike
and Roelof should put their
heads together. It sure would be sweet to have ALL those features in
one program. :) <sigh>" A swift rejoinder came
thus "Roelof and I are
collaborating on this already." Now I don't know how far
down the timeline something like that could happen but it would
certainly be a welcome convergence of Seti utilities. Information
and caching in one package - I believe there is buccal foaming
imminent. Perhaps I should mention in this advert for SetiDriver
and SETISpy
that one of the best but least reported features of these gems is
the near instant support available from Roelof
and Mike on the alt.sci.seti
newsgroup. Mention a problem or niggle and within an hour you've got
your answer. Bit like Dell but without the 'signed in blood'
techsupport contract.
Another reminder of how bios settings can make a huge difference
came from a regular newsgroup contributor Lawrence
Kirby who answered a query about slow PIII performance by
suggesting a check of the 'inorder queue depth' to make sure it was
set at 8 not 1. This is a memory transaction pipeline feature that
allows data to be queued and used far more efficiently than would
otherwise be the case. The data stream becomes a flow rather than
fast drip! The 800 Coppermine in question went from 9.5 to 5.5 hours
with that tweak on a BE6-II board. If only life was always that
simple.
Some results and I say 'some' as there are not many around these
days, what with questionables, hoax's, non-benchmark WU use, beta
submissions, genuine errors and the plain daft. I love my job...so
why am I doing this? For instance MAC owner Alexandre
certainly didn't take a few thousand hours to finish a WU so I'm
waiting on a revised number for that one. Chad
please get in touch as your T-bird time is bloody marvellous but do
I feel confident in telling the world before you give me some more
details, no way? My guess is you missed the first digit off the
submitted time! You might however be interested in the Thunderbird
overclocking article at Tweak3D
by way of light reading (5 pages with pretty dull pictures) while
waiting for RB's WU to crunch. I
can feel the hate mail in my box already.
On with the show! Regular bencher neo1999
has pumped a 200MMX to 280MHz (2.5x112) which is quite a feat in
itself and this hot little baby fills its nappy (diaper - okay) with
a 10:49. Way to go dude but your comments will have to wait until
the scanning public gets the new results page, ETA unknown to all
but RB & Knight,
plead with them, right now... Andreas
has a PIII running at 560MHz and the 2.4 Cli but can only croak out
a time of 8:23, sounds like tweak time down on the Mobo farm there,
bios first, RAM second. Unless you are using a laptop I'd ditch the
ramdisk too, it's a good discussion point in forums but a pointless
exercise in self-empowerment that has been chewed to death. Keep it
if you must but it really is a needless extra layer of complexity
that only real die hard DOS-heads claim significant benefits for. A
minute plus or minus is the more common experience which put against
the chance of losing the entire processing time due to a crash is
hardly worth it. Of course you might be the one person whose system
has never crashed, apologies. Ramdisks - love 'em, leave 'em or just
plain ignore them. More hate mail, can't wait.
Quite a goodly offering from gerdk
whose dual Dell Precision 410 modestly running at 500Mhz is a model
of Linux efficiency, kernel 2.2.16 should you need that level of
info. Two WU's completed every 7:52, great for the totals. The
occasional vanilla Athlon appears here in the form of betsy's
600 with only the standard bios and memory tweaks to speak of. A
6:51 puts this machine in about the right area, be pleased and
patient V3 will appear soon! I have this vague memory that Guy
Olinger named one of his systems Betsy
or maybe it was Betty in order
to differentiate the beasties that inhabited his domestic farm. Good
name and idea either way. Finally just a few minutes down the road neo1999
pops up again with the power of Celeron II (not), a 105 bus and 9
multiplier (ouch, though there are higher) conspire to reward with a
6:46. This 950MHz monster overclock is on a BE6-II and CAS 2 memory
so nowhere to go hardware wise. About as good as it gets with these
cycle-vampires. The previous generation Celerons with lower
multipliers gave more overclocking options and far better SETI
crunching times. We knew the new Celeries were not up to much
months ago, shame. Minor consolation is that it probably runs most
other apps very fast and it is the highest clocked Celeron in the
results to date. I can provide hankies and a shoulder too if
necessary. Right where did I put my medication, damn RB's
swiped my lithium...again.
Max out.
|
