ah
#🔌│tech
reef tundra
Broiler is just a very hot heating element very close to your food
cursive summit
Mine can't I think
Well mine has convection / microwave / grill
And I can use combinations of two of those
But not all at once I think?
reef tundra
Most likely, for power use reasons and fire safety
vagrant pagoda
convection oven is god send
cursive summit
Fam my oven is like 400 euros
I bought it for 150
But it's worth at least 300
They don't go for much higher than that
reef tundra
Buy more product
reef tundra
Ceramic lining? is it an makeshift forge aswell?
crisp granite
wut
you actually cook and eat turkey?
gross
I can't stand turkey, or poultry in general
funny, considering my grandfather on me mum's side was a chicken farmer
cool
good to know that spread to polynesia too
or is still alive and well as a tradition there
analog aspen
How hard does one had to press to make such a dent on the RVU?
Anyways mine had noticeable shining gloss just over a year of use 
(Picture is from Reddit, not mine)
analog aspen
Yeah idk never had that problem on any of my mouses that have side rubber
But my RVU have noticeable LMB & RMB shine
crisp granite
But my RVU have noticeable LMB & RMB shine
that's just skin oils collecting on the plastic
it happens
crisp granite
Shit only lasted 1 year?
I only think razer covers actual defective parts, not hard use and acts of god (sadge)
analog aspen
Nah it’s like ABS wear. No cleaning is getting it off
It’s permanent wear smoothing out the plastic surface
I just checked last night but forgot
One sec
Why tho? I used to use that on my old DA all the time
Not that bad
Actually the LMB and RMB shine/wear only happen to me on my RVU
DA, Zowie, Logitech, never had that problem
Yikes
8 years old DA. Good as new
Taiwan. Old v1 2019 batch
Week 44 so around October
Don’t think newer batch change anything about the shell tho
@crisp granite do you have the newer patch? I just found out they came with cool clicky click microswitch
crisp granite
I don't have a viper ultimate
crisp granite
viper mini
analog aspen
O
Pronin said Viper Mini received the update last
Check if you have the new switch
crisp granite
the gen twos?
analog aspen
Ye
crisp granite
I bought this thing not that long ago
analog aspen
I’m tempting to get a mini just to get the switch lmao
It’s going on sale for like $25 once in awhile
crisp granite
I'm like 350.3% sure I got the new switches
analog aspen
That’s an odd %
crisp granite
shut
analog aspen
crisp granite
but seriously
I've tried the old optical switches
these feel way better
gotta be the new ones
and personally, I like how these are less sharply clicky, like contact based mouse switches
misty saffron
meanwhile me with my $50 2.5 year old hyperx pulsefire surge 😎
actually a very good mouse
misty pier
i had a steelseries rival 100 that lasted me 3-4 years
i bought it for very cheap
my hands fit the viper mini perfectlyy
i dont need to do a claw grip or anything
analog aspen
reef tundra
You just want to kill the mouse? cool
analog aspen
Well I’m on the same boat. Considering the Keris but ngl spec is pretty bad. Viper getting oil corrosion is bad in my book (thinking 8kHz). Only GPX is left
choke grip
Keris is around the size of model d
analog aspen
Yeah I don’t think you can even finger grip Viper mini properly
RVU is already pretty small for me
analog aspen
i dont need to do a claw grip or anything
That means it’s big enough for you to not have to claw grip. Clawing is when the mouse is a bit too small for you. Fingergrip is when it’s really small for you
Buy an EC1-B
misty pier
small hand gang
crisp granite
https://videocardz.com/newz/modder-puts-16gb-memory-on-geforce-rtx-3070-and-it-w...
hm
I have a 3070
and enough vodka to understand what he's saying
astral haven
there are people old enough to be on discord who were born after wifi n was released
misty saffron
THERE ARE PEOPLE WHO DONT KNOW THERE IS WIRED INTERNET MAN
crimson wigeon
Yikes
misty saffron
i’m so glad i wasn’t born later
zenith sonnet
Wow
Second dishwasher was delivered today
Similar damage then the first one...
Is it that hard to handle a dishwasher??
hard thicket
apparently yes ?
fair gull
xea will never get a dish wisher
zenith sonnet
It's not a EUR pallet
It's just 5 boards stabled together
Still, a EUR pallet would probably not have saved it
Looks like impact damage from falling/sliding off
It almost slid off the 'pallet'
Well, the wife is gonna be upset about this one when she wakes up, that's a given 
reef patrol
old marlin
@fair gull 
where do you host websites?
pale sigil
@old marlin @old marlin on his server
ashen spindle
psure its on netcup
fair gull
<@130036426912825344> <a:Pepethink:619039042561900544> where do you host website...
netcup. used to use their webhosting but after a dataloss debacle i use the root i get from them
id only recommend the webhosting packs with the caviat of not transferring domains to netcup that are already configured on netcups plesk panel
old marlin
Yea seems like we will end up getting the domain via netcup
Strato is absolute 
for what we want lol
fair gull
ngl netcup isnt always the cheapest but they offer solid products and solutions
the support is quick altho not all issues get resolved in the interest of the customer
old marlin
Yeye
old marlin
We'll host ourselves 
fair gull
good idea
old marlin
Strato is like "buy 17 different things to get the domain"
Like wtf
Even at mcdonalds i can buy a burger without a menu
halcyon moss
i host on tilaa
and personal one vps i have on transip
but dutch, so.. yeah, maybe you want german
ashen spindle
clearly they ran off with it and sold it for profit
pale sigil
"Shipped From EVGA: Pending"
nvm
damn, taking their sweet time
😄
@ashen spindle 
as long as its done before next credit month its fine but i dont like using my credit card and i had no choiced
reef patrol
You folk and your credit 😄
ashen spindle
credit is too american for me to ever do
pale sigil
Well
ashen spindle
wait no I guess technically I did use credit when I did crypto trading 
pale sigil
It doesn't say which PSU is being send out
thats changed
xD
willprob be sent today or tomoz
since friday wont really count as a business day
usually start the next week
i mean it is a business day but you get me
vagrant pagoda
I know there is a psu shortage, bit I've only ever bought silverstone psus and never had one fail on 10+ years
Idk who makes evga's psus
astral haven
Everyone sources all sorts of stuff
vagrant pagoda
But I don't like buying from third parties oems
astral haven
Neither Silverstone nor evga are oems
vagrant pagoda
Silverstone dose in house design and manufacturer for thier psus...
vagrant pagoda
I thought they were a subsidiary of the oe
ashen spindle
seasonic has worked great for me so far, probably my go to for psus
vagrant pagoda
Who am I thinking of?
astral haven
Seasonic is the only one
Fsp is getting into retail now tho
Seasonic is just bad for overclocking
ashen spindle
how so?
vagrant pagoda
Yeah I'm der8auer runs seasonic on all his oc stuff
They have been his sponsor for years
astral haven
Most brands you want 20% headroom, seasonic you need 30%
ashen spindle
hmm that's the opposite of what I remember, what I looked at most of them had decent headroom
vagrant pagoda
So they have working OCP, and that's a bad thing?
astral haven
Yea well if you have an unlimited budget lol whatever
No
Ocp shouldn't trigger when max current is less than psu wattage
ashen spindle
nono I mean in terms of what they can take for peaks, let me see if I can find the video I have that from
vagrant pagoda
I can't see any data to back that up
astral haven
Yea everything is fine on paper with seasonic, except for some extra ripple on 3.3 and 5v
vagrant pagoda
Considering they do over rated in jonnygurus hotbox test
astral haven
Their Ocp points are correct
astral haven
But they just can't run shunted nvidia cards unless they are heavily overspecced
vagrant pagoda
That dosent make sense
If they can do rated in the 12v only test
For 10+ minutes
They will pass any bench
vagrant pagoda
What gpu bench is an hour?
reef patrol
🤨
astral haven
Short spikes are alot more stress than long term power usage
ashen spindle
I think it was a computex/ces video from either steve or roman where they were at a psu booth and topic was seasonic and how they usualy have like 30% headroom above rating, granted I dont think it was mentioned for how long
vagrant pagoda
Also the hot box test is 100% rated for 10 minutes for just one test
ashen spindle
and not really easy to find since it wasnt the main focus of the video...
astral haven
Yes their opp is good
vagrant pagoda
And the ambient temp is 50c
astral haven
Look if you aren't overclocking it doesn't apply to you
Unless you are getting a 3090
ashen spindle
I think the focus is more on short spikes that occur especially on OCd hardware that only go above rating for a minuscule timeframe, 100% load test doesnt matter there
astral haven
Yup
fair gull
but dutch, so.. yeah, maybe you want german
damn those vps are expensive
ashen spindle
but I would still assume if it can handle like 30% above rating, it could also deal with short spikes that dont even go that high?
astral haven
The problem is the rising and falling edge of the spikes
You can think of it like whiplash
vagrant pagoda
Look if you aren't overclocking it doesn't apply to you
I ran a pair of 450w 290xs for my daily rig for ~3 years on a chiller
astral haven
Electrons have momentum
vagrant pagoda
I know what I'm talking about
astral haven
290x aren't that spikey
fickle ore
@astral haven How much m/s?
vagrant pagoda
Lol wut?
fair gull
@halcyon moss the same spec vps with less traffic is 20eur a month on transip
vagrant pagoda
That was a dual psu rig
astral haven
295x2s are spikes asf
vagrant pagoda
And I had to run a 1200w for just the gpus
astral haven
But 290x had caps
astral haven
Look
ashen spindle
but the card itself might not spike as hard
astral haven
I can tell you the facts of the matter
ashen spindle
higher tdp does not instantly relate to higher current spikes
unless we talk about same VRM etc.
vagrant pagoda
Show me some data to back it up
astral haven
Idk how many systems I've helped troubleshoot?
ashen spindle
I mean, idk if anyone really did load analysis of a 290 card
astral haven
Tomshardware did
astral haven
Hard to mess up an oscilloscope reading
Ok actually that'd not true it's pretty easy
.com
It was a while afo
ashen spindle
hmm guess that means it's still possible it was igor who did it then
as he's doing it now on his own page
astral haven
Very not good for a trigger happy ocp
Even my corsair psu quit on me and corsair is known for being very precise with their ocp
If that's a good thing that's up to you :P
fickle ore
This only matters if you load up ur gpu to 100% and whos kidding here we all play lowest setting for competitive advantage anyway
astral haven
Nah
ashen spindle
yeah, I too prefer to enjoy witcher 3 on lowest settings on my 3090 
fickle ore
xD
astral haven
Most spikey power is when the clock is highest
Low settings high fps is really bad for psus lel
Well
fickle ore
yes
astral haven
If your gpu load is for example 10% that means that 90% of the time the gpu is drawing idle power
fickle ore
but your gpu isn't gonna use 1kw for 1ms if you're playing on optimized XD
astral haven
The moment your cpu provides it a frame it kicks into gear and spits out the frame asap
If you have a gpu that has bad coil whine you can literally prove it
Take furmark and make the window smaller and smaller
vagrant pagoda
I meant that seasonic psus trip under thier rated power
zenith sonnet
Your gpu will screeeeeam
Mine doesn't 
astral haven
If you have a gpu that has bad coil whine you can literally prove it
I said if
vagrant pagoda
I have luckily never have coil whine
fickle ore
@astral haven if it worked like that my GPU in Valorant would scream, because my avg framerate is 900. But no, it rather screams in PoE where my avg framerate is 120.
Also Blackdesert Online launcher sucks
vagrant pagoda
Only issue I've had is my 7970s and 290x had to low of a 2d clock
pale sigil
I see black desert
vagrant pagoda
And would fuck up visuals on desktop with eyefinity
pale sigil
You grabbed my attention
ashen spindle
I get coil whine whenever I start steam vr 
bc my 6800 renders that at like 600fps
vagrant pagoda
So I just flashed bios to raise them
pale sigil
zenith sonnet
I need to get it up to about 5000 FPS to be audible
fickle ore
It's 100% based on the amount of GPU power the gpu thinks the game needs. In Valorant at 900 (cpu bottleneck) it doesnt even care because it barely loads like 30% of the gpu cores prob -> lower current
astral haven
With an unlocked bios my 780ti was so screamy that I could pretty much use it as a speaker lel
vagrant pagoda
My 1080ti, v and 64 don't have any issues that aren't driver
ashen spindle
also gotta keep in mind both loadspikes and coil whine can vary drastically on the same SKU depending on what model you have
fickle ore
yes, so % load of teh gpu cores
pale sigil
@fickle ore launcher used to be worse
fickle ore
or transistors if you will
vagrant pagoda
Also I'm glad actual vrm heat sinks came back into style on motherboards
I always hated the dumb extruded aluminum look
ashen spindle
they are not back bc of style they are back bc it's needed on some 
vagrant pagoda
I just want the day of simple motherboards back
ashen spindle
yeah that train is gone and not coming back
vagrant pagoda
Yeah
astral haven
Z77 not z78 oops
vagrant pagoda
My k9a2 platinum is my favorite mobo of all time
It had everything
Back when a 5 phase ran a 105w cpu....
astral haven
Peeps now running a 150w cpu on a 4 phase
astral haven
Also back then if your cpu drew over tdp you got crucified
vagrant pagoda
Yep
astral haven
No doge it's pretty normal
crisp granite
most of us are reasonable and aren't running Ryzen 9s off of A series mobos
vagrant pagoda
That was my last msi board
ashen spindle
gimme 4 phases for my W3175X 
vagrant pagoda
I upgraded that to a gig 990fx ud9
halcyon moss
<@76617282586947584> the same spec vps with less traffic is 20eur a month on tra...
transip is not cheap, would not recommend tbf
astral haven
3 phases usually teamed but still 3
vagrant pagoda
Which was chilled loop setup v1
I kept blowing vrms
The amazing thing about that board
It would blow a vrm phase
And reboot
And work
I have the vrm block still
And it has like 4~5 skid marks from blown mosfets
vagrant pagoda
I would turn off the oc
And wait for replacement board
For a week or so
Missing a vrm
I think I was hitting pcb limits
astral haven
Fx
vagrant pagoda
1090t
It was probably burning some internal fiberglass layers and shorting the power plane
I had a chiller loop setup with a few point controller
astral haven
Phenom on 990fx???
astral haven
??
vagrant pagoda
That was the main use for them
astral haven
8120 not 8130
vagrant pagoda
Especially for oc
vagrant pagoda
I was running my 1090t at 4.2
1.55v
She was cooking
That rig would ocp a 1200e
fair gull
transip is not cheap, would not recommend tbf
tilaa is equally expensive damn
glad netcup has an eu website
vagrant pagoda
Which was the biggest you could get then
So I have a 800w for everything other than gpu
And the 1200w for the gpu
wheat plume
I hate hw they call it a payload xd
ashen spindle
.. thats a normal naming for it?
wheat plume
perhaps, I just always associate it with malware
ashen spindle
not really their fault though
wheat plume
spyder kept crashing when I installed it
so now I installed anaconda cause the spyder website says that is the best way to install spyder
and now wtf I do?
vagrant pagoda
Payload is the correct term, you just don't do a lot of remote deployment stuff
wheat plume
hmm
and probably the most unique example of the mounting possibilities.......questions what he just said
Arun - 2021
wheat plume
:')
smol af camera you can mount to anything
yes but youd not wear your phone on your helmet if you wanted to record a bike ride or whatever
:P
crisp granite
hm
I could use this for sailing...assuming it's waterproof and class rules allow it or say absolutely nothing about it
wheat plume
it edits for you...WOT?!!
it can auto sync video clips to music
Yes, its waterproof
wheat plume
the what
crisp granite
it's a legal document of what you can and can't do for boats in a certain class
for racing
wheat plume
ooohhh
crisp granite
it keeps the competition fair
so, I can't use an ultralight hull that weighs 90 kg less than a normal one and just smoke out the competition
or a larger mast and sail combo, different foil shapes
stuff like that
wheat plume
oo, its $300
crisp granite
some class rules have rules about nav aids, mainly larger boats
this might classify as one
wheat plume
vagrant pagoda
Any of y'all have a aio with a screen?
crisp granite
https://youtu.be/O_XhRdWdphU
I was gonna get one
but 400 yankee doubloons for one with split backspace and iso, in the industrial grey...
no go for me
wheat plume
it is pricey af...
wheat plume
its chonky too
crisp granite
mhm
fair gull
lead weights
crisp granite
totally forgot what the case is made from
quartz solstice
im watching it while im in class LGR is awesome
quartz solstice
that space bar tho getting stuck
crisp granite
that's an issue with the stabs
my bad
it could be
the springs need to be set just right
it sounds...
delicious
me must get one
wheat plume
is so expensive ;-; you better really love them :P
crisp granite
gonna swallow the 403 doubloon plus shipping to canada
at least I get UK ISO and HHKB split right shift and split backspace for the money
wheat plume
no function key row
fair gull
wheat plume
hmmmm
fair gull
when u restore files from a drive with 200x json decode speeds of gta5
old marlin
"RJN" 
analog aspen
Sucks that 2 best mouses on the market’s shapes suck for me
I’ll probably need longer mouse/mouse with hump for my long hands
old marlin
^
if you got somewhat large hands you will almost certainly be satisfied with either a burst pro or XM1r
chrome estuary
I like my MX Ergo personally
analog aspen
if you got somewhat large hands you will almost certainly be satisfied with eith...
i got long hands but rather narrow... 18.7cm length, 8.5cm width (from thumb nail) or 10cm width (from furthest points)
looking into those now...
I can only claw grip the DA xd
analog aspen
width ye
analog aspen
if measure like this then 10cm
old marlin
ah
analog aspen
if from the thumb nail then 8.5 cm
old marlin
yea ok nvm then
you just have an proportionally large sized thumb
and proportionally small sized rest of the fingers
zenith sonnet
I also wanna look at new mice. What would fit a 11x20cm hand?
wheat plume
my thumb is slightly, barely longer than yours, but we have the same hand shape
analog aspen
partly because of camera angle
wheat plume
no wonder it was always too loud or too quite .-.
when playing music on gnome desktop
I want mine to be 2% like on windows
much nicer than fucking 6% .-.
o.o
you volume control per app?
ooooohhhhh
my headphones dont have a volume dial xd
my speakers do though
huh?
wheat plume
im lost
vagrant pagoda
Eh I set the level on my speakers and fine tune with digital control
Pc is connected via hdmi digital audio
I just change the digital level to normalize it
I very rarely have to adjust the amp level
Nah
Not really a need
You ears have enough dynamic range for it
With headphones it's bad
With speakers it's ok
It's not that loud
I don't have it that loud
The speakers have great sensitivity
It's not as bad with good speakers
I can hear a whisper as good as clipping
And I have clipping with 100% digital set lower than the most I can tolerate
So it can never get too loud
I have a big set of floor standers
Dual 8" mains, 4 inch mid, and a 1inch tweeter
With preamp crossover so they are bi amped
Then what about everyother type of media?
analog aspen
anybody in NA had experience buying from them?
I wanna try out the burst pro but gonna get fucked if they dont accept return
amazon OOS
obviously
that's fair tho
~$5 for 15-30 days testing
my problem is if they won't accept return if nothing is wrong with the mouse
unique quail
my keyboard aint workin no more
[12:02 PM]
when i switch modes it doesnt register wasd movement
[12:02 PM]
on the 360 degree movement mode
[12:03 PM]
this is for fortnite btw
analog aspen
@unique quail #archived_helpdesk_for_keyboards
unique quail
ty
vagrant pagoda
Maybe in browser?
crimson wigeon
Oh joy
reef tundra
Monopolies are great.. for the business
thick elk
For replaygain I generally parse my collection using loudgain How you enable it depends on your music player.
#!/bin/bash
# add all files into an array
echo "parsing audio"
readarray -d '' filedirs < <(find . -maxdepth 2 -mindepth 2 -type d -print0)
# running Loudgain (rgbpm) using gnu parallel
echo "Applying REPLAYGAIN"
parallel --bar rgbpm ::: "${filedirs[@]}"
Basically the script I run
proud nacelle
realisticly speaking, how important are updates? because it seems like I can run everything fine without updates (Win 7). Same goes for software. I get that updates exist for bug fixes and security fixes... but if ain't those two, then what's the point?
ashen spindle
idk, you're being way too generic about them
ashen spindle
bug fix, security fixes, compatibility, new features
proud nacelle
I suppose. But to me, new features is actually a hinderance. Compatibility, not sure what you mean, but I would like to find out.
ashen spindle
as in making an update so the new software x can work on your system
for example raytracing needs a certain version before it can work due to API changes etc.
proud nacelle
Well
I don't use raytracing so I can't have any say on that. But I mean, running Win 7 with no updates and... stuff is working fine.
Just makes me wonder.
ashen spindle
well yeah if it would break down on its own there would've been an urgent patch for it
proud nacelle
It would be nice if I knew what exactly those updates did
Or if I could inspect them
ashen spindle
unless there is a big security or performance issue I just run some time behind so if there are bugs they will be found and then I update
dont really care what it is at that point
cba at points figuring out if I need to update to fix any issues I get potentially, I just do it before I run into that
proud nacelle
I mean yeah that's what I do for the most part anyway
wheat plume
its blue and yellow
proud nacelle
real quick dumb question: anyone else getting a cert error at https://dotnet.microsoft.com/ ?
crimson wigeon
Just you
ashen spindle
wheat plume
dmenu is neat
reef tundra
I know right
reef tundra
You can style it however you want, code compile go
wheat plume
I saw someone make rofi blur the surrounding background when open
was very neat
oh, hmmm I see
reef tundra
Probably can do the same with some xorg generic blurrer
reef tundra
picom or something
wheat plume
and also some compositors too
ashen spindle
either mindfactory was told to stop the sale, or 250ish was the entire stock they got for the 11700k
reef patrol
The latter most likely.
shadow minnow
Or both
ashen spindle
yeah it's the stock ran out, at leas the one they could sell
but that either means the nda break happend at only a smaller supplier, or there is even less stock of intel cpus than zen3 lmao
shadow minnow
It’s 14nm so I think the first
ashen spindle
imagine making a worse cpu, that costs more to make and has less revenue, to also produce less
shadow minnow
But... but our real life usecase benchmarks show how intel is better
ashen spindle
im so sorry for the poor engineers who had to backport this for basically performance regression compared what they had
also a bit sad that the laptop processors are a gen ahead of both desktop and server now from then
ashen spindle
no 10nm is just broken
proper crystal
Server is still coasting on Skylake. Though Sunny Cove based server parts reportedly shipped 100k units as seeding before official launch.
shadow minnow
It’s actually apparently as good as TSMC 7nm, just the fact that when we will se it on desktops, TSMC will already be at 5
ashen spindle
yeah but pre launch doesnt really count
proper crystal
Some versions of 10nm were broken. Seems Intel fixed it enough for themselves.
proper crystal
yeah but pre launch doesnt really count
It is still 100k units. Of big server chips made on newest version of 10nm.
ashen spindle
they "launched" 10nm in 2018
proper crystal
Yes, but nodes evolve. Intel's 10nm especially
ashen spindle
that time they had cannon lake laptops and claimed 10nm is in production
just to not fail investor goals lmao
now 3 years later 10nm barely scrapes by and 10nm seems to be decent-ish
proper crystal
The bigger issue for Intel is that, normalized for node, their Cove microarchitecture which has PPC parity with Zen3 takes up 1.25x the area.
shadow minnow
they "launched" 10nm in 2018
that’s now normal, They launched nvidia ampere and new amd GPUs 2020 and yet
ashen spindle
insane demand != no supply
proper crystal
Again, they already shipped 100k massive server chips made on 10nm.
proper crystal
that’s now normal, They launched nvidia ampere and new amd GPUs 2020 and yet <:...
Both consoles and new GPUs sold crazy amounts of units. The supply couldn't really be higher, it's the insane gamer demand that causes the issues.
ashen spindle
server chips yes, but I doubt all or even many of them are the fully enabled SKUs
proper crystal
server chips yes, but I doubt all or even many of them are the fully enabled SKU...
Yeah, but still, it's progress. Part of the issue is that Intel doesn't have a competitive interconnect. Large dies on N7 also end up with a lot of defects and partially disabled dies - although N7 is way more mature than 10nm Intel.
ashen spindle
sure but if lets say 80k of those are 6 core xeons, that is still a broken node in my eyes lmao
the number alone means little
proper crystal
sure but if lets say 80k of those are 6 core xeons, that is still a broken node ...
They won't be 6 cores. That would be suicide for Intel.
I'm talking about Ice Lake SP.
ashen spindle
I know, cascade lake went down as far as 6 iirc
I doubt it is that bad, just saying 100k is very deceptive potentially
for all we know they might not even have final specs, not work at planned TDP etc.
proper crystal
Cascade also launched when mainstream Intel was 4 core or 6 core.
proper crystal
for all we know they might not even have final specs, not work at planned TDP et...
They do, almost certainly. This isn't prototype ES seeding, this is full production units seeding.
ashen spindle
ok if we go to cooper lake which is the latest that also goes down to 8..
proper crystal
Cooper is just plain weird. Anyway, 80% of the seeding units being 10 core or less would be suicide.
ashen spindle
im just saying, usually nobody discloses these numbers, intel discolsing it when everything else points towards bad things doesn't make me believe the number is anything but deception
well yeah that's a bit extreme, just saying it could be and the number would still be 100k
proper crystal
Either way, Ice Lake SP is catching up in single core perf to Rome, with fewer cares on top SKUs and with Milan launching sooner, so...
ashen spindle
- based on anandtechs article 1600 wafers could fulfill the 115k dies
trying to find the numbers of waferthroughput of tsmc 7nm but that doesnt sound great
yeah, suddenly that 110k chips is a tiny ammount
https://www.anandtech.com/show/16539/intel-ice-lake-xeon-scalable-shipments-to-date-30-customers-100k-units update at the bottom if you want to see the math for 1.6k wafers @proper crystal
proper crystal
If it was only 1.6k wafers, then 10nm would have to yield pretty well.
It is a tiny amount for full production, it isn't tiny for a seeding run.
ashen spindle
uhh that statement only holds up if we know they only had it running for a short time
" Intel originally stated that it was sampling Ice Lake Xeon with key customers as far back as May 2019. With high volume manufacturing starting in Q4 "
so almost 2 years
ofc they dont run at full capacity so you can't directly compare it to regular production with 4k+ wafers a month at tsmc
at the same time intel is much bigger and their own numbers would be even higher
proper crystal
Either way, their current 10nm node is good enough for them to use it on a big server die.
ashen spindle
idk I remain sceptical
10nm SF? sure that seems to be running fine
but ice is on regular
or 10nm+ rather I guess?
fickle ore
11700k vs 5800x core-to-core latencies btw
ashen spindle
apple pays more and thus it costs more
pay more and you get any hardware atm
so no real difference
shadow minnow
Well, you have to pay lot more for other hardware atm, not just more
shadow minnow
11700k vs 5800x core-to-core latencies btw
Although it is really bad for intel, afaik saw somewhere this graph is not very correct. not sure if the second cpu is not 5950/5900 and also there are yellow numbers all around afaik. But still, the intel loses badly
ashen spindle
what
those are both numbers from anandtech
and you wont find many people in journalism doing the same or better latency stuff than andrei
shadow minnow
proper crystal
Although it is really bad for intel, afaik saw somewhere this graph is not very ...
It is 8 core against 8 core.the scale for colours seems consistent as 5800X does not go above 20ns core to core.
shadow minnow
Either 5 5800 and 5950x have same latencies or there is sth wrong with that graph afaik
ashen spindle
.. you do remember ryzen has chiplets yes?
of course they have the same
if you just look at latency withing the 8 cores on one chiplet
proper crystal
Either 5 5800 and 5950x have same latencies or there is sth wrong with that grap...
5950X is from two CCXs - that's why you have the jump when going from one cluster of 8 cores to the other.
ashen spindle
you have all the info about a 5800 in a 5950, like you get a 5600 data from a 5900 graph
proper crystal
In Zen1 and Zen2 CPUs it'd be clusters of 4.
shadow minnow
If it’s completely same latencies, then yes I guess, just that this is basically cutted out of 5950x graph, but not really sure how is it completely with latencies tbh
shadow minnow
Does not look completely same, first quarter is little bit different to forth, although the differences are negligible
proper crystal
The latencies will vary a little, which you even see if you compare the 5800X and 5950X - tiny variations are inevitable.
Other than that I don't understand what you mean.
proper crystal
Does not look completely same, first quarter is little bit different to forth, a...
Yes, tiny variations.
ashen spindle
yes because not every piece of silicon is identical, but thats the only difference you get
proper crystal
It's more about random variation than silicon level stuff. But that's a technicality.
proper crystal
it's in both cases a look at one CCD
CCD isn't a useful term here. CCX is.
ashen spindle
core 0 cares as much about core16 when sending to core 1 as it cares about what gpu the pc has
eh I guess, since they are just identical in this case but not always lel
proper crystal
The fact on Zen3 one CCD contains one CCX doesn't mean the parity will always be there.
CCD is a physical term, CCX an architectural one.
ashen spindle
either way, tomorrow we get more data from hardwareluxx most likely, they also bought one
though I dont think they test latency as intensive
astral haven
the real question is memory latency
cause obviously the ultrawide ice lake cores are crying with jdec ram
proper crystal
the real question is memory latency
The real question is measurable end performance. And there RKL seems to fall flat on its face.
astral haven
with slow ram sure
proper crystal
cause obviously the ultrawide ice lake cores are crying with jdec ram
Then Intel should've made the caches larger or work on caching policy and prefetcher.
ashen spindle
you're asking alot for a backport, it's not a new uArch besides backporting basically
astral haven
^
proper crystal
Slow RAM? It's what Intel certified the design for. All tests done by Anandtech are at max spec of the CPU - including Zen3 and Comet Lake tests.
astral haven
this was designed to run with lpddr which is waaaaaaay higher bandwidth than ddr4
no
jdec 3200 != high speed 3200
proper crystal
you're asking alot for a backport, it's not a new uArch besides backporting basi...
Oh I know. But if Sunny and Cypress Cove are so sensitive to memory latency, then the core Sunny Cove design is unbalanced.
ashen spindle
willow cove was designed for 10nm, it was then backported to 14nm to have a working product, part of that means you have suboptimal designs for memory controller etc.
astral haven
literal 40% performance with fixed timings
proper crystal
willow cove was designed for 10nm, it was then backported to 14nm to have a work...
Cypress is backport of Sunny, not Willow.
astral haven
and here are fps values
ashen spindle
nope, its ice lake backport, not tiger lake
proper crystal
literal 40% performance with fixed timings
Yet all tests done by Anandtech run on the best JEDEC spec memory guaranteed by either Intel or AMD.
astral haven
no?
Jdec of the speed isnt equal to the best speed of that memory speed
i mean the 5800x is in stock lel
proper crystal
Jdec of the speed isnt equal to the best speed of that memory speed
I know, but Intel and AMD guarantee JEDEC specs.
So 2666 JEDEC, 3200 JEDEC etc.
astral haven
nope
ashen spindle
intel had its own shortages that are still not over
astral haven
intel at least guarantees max theoretical speed on their rated clock
idk about amd
remember different boards will treat jdec differently
tuned subs can improve performance by 30% entirely without anything visible to anyone
proper crystal
Anandtech always tests on JEDECs. So their tests are comparable to each other.
astral haven
their tests are comparable to each other
but what happens if you actually feed the cores?
thats what is interesting about rkl
its for the 1% of the 1%, but its still interesting
vagrant pagoda
People never understand how much ram effects performance
proper crystal
Oh, then OK. A lot of people do complain about testing at JEDEC in general, I thought you're one of those.
astral haven
oh no at did the right thing
but i want overclocking
so im gonna have to wait a bit 😦
i mean
proper crystal
Also Intel fanboys are using this to say that RKL actually will be a beast or something.
vagrant pagoda
Going from base settings to 4000 cl16 on my ryzen 5 setup is almost 40% in some games
astral haven
well its actually insanely fast
proper crystal
well its actually insanely fast
Doesn't seem to be compared to other contemporary designs.
proper crystal
Eh. Geekbench isn't a very good benchmark. I also expect 11900K to be very much a paper halo - same situation like 10900K and 11850K.
astral haven
whats interesting to note is this
53ns
anandtech got 71ns
if comet lake bioses can do 53ns already, theres not much distance to make up to hit comet lake performance
and if it can do comet lake singlethread with the same anandtech multithread scaling? that would be fantastic
proper crystal
Anandtech is more trustworthy IMO because they test various cases. Benchmarks usually run one case, and also CPUs do detect big benchmarks and alter their behaviour accordingly.
astral haven
yes but ram latency is ram latency
proper crystal
Not just RAM.
proper crystal
You also have the internal mem controller setup and coherency system, and prefetcher. Different data types and sizes will give out different results.
astral haven
and there are plenty of benches online with people in the 40-50 range
ashen spindle
For all we know intel might’ve crippled the 11700k to have reasons for selling a 11900k lmao
proper crystal
For all we know intel might’ve crippled the 11700k to have reasons for selling a...
I doubt they can make enough 11900K for that xD
astral haven
11700k isnt that much slower
and intel can make a tooooooon of chips
70 billion dollars makes things happen lel
analog aspen
For all we know intel might’ve crippled the 11700k to have reasons for selling a...
Waiting for Ryzen this Gen and you showing that it’s better than Intel again 
proper crystal
and there are plenty of benches online with people in the 40-50 range
Plenty of benchmarks that use single or two data structure types to gauge latency, and benchmarks that are often detected by the CPU which then triggers "optimizations" that only happen for that benchmark.
ashen spindle
So much they basically stopped the 10900k because they couldn’t bin enough and made the 10850k a thing yea
astral haven
ryzen isnt releasing any more consumer cpus on ddr4
at least not any ones faster than the 5950x/5800x
karbin
please
proper crystal
So much they basically stopped the 10900k because they couldn’t bin enough and m...
Exactly. I can't wait for 11850K :)
astral haven
if you dont understand how memory ladders are made, dont try to pretend you do
theres only one way to measure
and everyone measures the same way
there are different ways to present the results sure
thats why anandtech says 84.5ns in their graph when their actual result was 72.7ns
84.5 isnt wrong its just not the normal way that everyone else reports it
and tbh i wish everyone did use total latency but 🤷♂️
analog aspen
I didn’t win anyways. What’s to expect
Also the chips look shit
At this point I’m happy I got the 2080Ti when I did despite people memeing
Now I have performance card rivals 3070 while people pay $300 more
ashen spindle
Wtf how cheap did you get your card for
astral haven
wut
analog aspen
Corona, stock hit, people waited, chip shortage, tax war
astral haven
ok but guess how many 3080s showed up on the steam hardware survey
aka cards sold to gamers
analog aspen
Because this gen’s chip was better that people are now chewing off each other’s to get it
astral haven
yea
analog aspen
astral haven
but gotta understand that alot of people got their chips
analog aspen
but gotta understand that alot of people got their chips
Was it worth all the hassle tho
astral haven
again try to guess how many people got 3080s
on steam
1080ti went up iirc?
oh no
it did go down
went down by ~ 20k
proper crystal
and everyone measures the same way
Maybe most common benchmarks do use the same data structures. Does Anandtech do so as well? Not sure, don't think so, and even if they did, it's not as likely that the CPU would detect their in-house latency benchmark. https://link.medium.com/Y3fLmc1Zueb
astral haven
they all use the exact same methods to get the latency
i forget the exact series of x86 assembler but all the benchmarks do the same thing
ok the 1080ti went down by 30k
actually maybe closer to 40k
proper crystal
Again, maybe they all do. But given the discrepancy with AnandTech, I'd guess they do something a bit different, or maybe they use both the traditional test and some more custom ones.
astral haven
no?
the difference is exactly what youd expect for jdec vs xmp
oh yikes
yup
its also a shame anand himself retired too
guy was a legend
proper crystal
the difference is exactly what youd expect for jdec vs xmp
So this is just about DRAM latency, not memory access latency?
astral haven
yup
i calculated dram latency from ats numbers
at got 84.5ns memory access latency
which at 11.9ish l3 turns into 72ish ns dram
vs the 52-38 that we see with xmp
and we do know that gb4 and aida64 calculate dram latency the same way so their results are directly comparable
proper crystal
Oh, then sure I guess. But memory access latency is what counts in the end.
Given how fucked Cypress Cove L3 seems to be, I don't hold out as much hope for a big improvement from memory timings.
Would expect more with higher DRAM speeds since those should also make the uncore clock higher.
astral haven
3733 dram clock is the highest we have seen here
and they hit sub 40ns
best ive seen on zen 3 is 48ns
proper crystal
We'll see how the scaling will work out. I think memory improvements will get choked by L3 and ring bus.
astral haven
and 2000 fclk > 1866
nah
intel doesnt have a ton of l3 but its as fast as zen 3s monster cache
proper crystal
2000 fclk on Zen3 is really high.
proper crystal
Well, it is slower than Comet Lake and other previous Skylakes.
astral haven
its not practical to hit that in reality where 39ns on rocket only took a day lel
yea sky can hit 25
but well
sky aint got the cache bandwidth
like look at that l1
its double the bw per core of zen 2/3
proper crystal
Well yeah. They have big L1 in Coves. But what really matters is the end performance.
proper crystal
I care much more about the underlying uarch. And Cypress Cove is a bastard child of an already shitty design.
astral haven
nah
ice is fantastic design
sure the l2 is a bit anemic but its fine
tiger is literally just ice with more l2
remember rocket is faster than the 3990x and 3995wx in some workloads
proper crystal
It's too fat.
proper crystal
remember rocket is faster than the 3990x and 3995wx in some workloads
You mean ideal case AVX512 stuff?
astral haven
?
yes
this is 220mm2 of die area beating out 600mm2 of die area
on a node 1/3 the density
proper crystal
?
About 1.25x the area compared to Zen3 (when controlled for process as much as you can).
AVX512 is a benefit for very limited set of workloads, which exist between CPU workloads and obvious GPGPU candidates.
It is nifty, but I'd rather have it as, say, an optional chiplet.
astral haven
1.25x the area for 6x the performance?
proper crystal
...in very limited set of ideal AVX512 applications...
astral haven
and 1.25x the area on a node with 1/3 the density
14++ is 35mt/mm2 and zen 3 is on a 95/105mt/mm2 node
proper crystal
"normalized for process" - it's core plus L2, based on Sunny, not Cypress.
proper crystal
You have exact measurements for Cypress?
If so, they managed to have the same area for Cypress on 14nm as they do for Sunny on 10nm. Now that's quite some magic.
I'm counting core + L2 cache, not the L3 since that can vary more.
proper crystal
Ideally I would count only the core, but it is troublesome to say definitively the extent of L2.
proper crystal
I know, but I'm talking about the logic itself.
astral haven
not including the literal most important part of the core (the caches) would be outright dumb
also we dont know the logic density of any of these nodes
proper crystal
And L3 and L2 are somewhat easier to scale than it is to change the logic core.
astral haven
only the geomean density between sram and logic
proper crystal
I know that. That's why I say "about 1.25x" and "controlled for process as much as you can".
astral haven
for ease of calculation i include anything on the ccd for zen and the equivalent for intel
because we know the exact transistor counts of the ccds we can extrapolate pretty well
proper crystal
Yes, and yet again, we are doing two different types of comparisons 🤣
astral haven
but the tldr is that sunny is very very high performance per transistor
power is dumb and intel is dumb for not taking full advantage of the fpus that they added on for avx512 (good examples are tigerlake and skyx)
but its still not entirely wasted
proper crystal
We'll see how Alder Lake and Zen4 will do. First gens of DDR tend to be expensive and slow though.
astral haven
d5 isnt gonna be slow
its gonna zoom
but oooh boy is it gonna be expensive
and i predict that some are gonna explode
😬
10900kf for 320 is 👌 tho
reef tundra
astral haven
tbh my platform moneh is also pretty available
so much so that im considering buying this
it would delay me until at least zen 4 in 22 but like
prettay
ew prebuilt
im opposed to prebuilts that cost more than the parts
reef tundra
Convenience is also worth something
proper crystal
d5 isnt gonna be slow
Slow compared to what is available with 2nd gen of CPUs on that memory.
vagrant pagoda
I still wish you could get gpus smh
I bought my 1080ti on release day and it wasn't bad
Now it's nuts
reef tundra
oldegg
That would be the middle of the average lifespan
They saved you from owning one
Its Japanese
vagrant pagoda
Dude
I have a first gen tundra
Same running gear as a brand new 4runner lol
Thing with the 4runners
If you ever want to lift it
Spend the money in the trdpro
All the other one have active anti roll systems
Which don't work right lifted
So most shops just disable it
The pro still gets normal sway bars
Eh
Manual off-road is meh
reef tundra
Manual gang
vagrant pagoda
Which is why they added crawl gear to the bronco
I mean my daily is a manual 4x4
And I always worry about clutch when I off road lol
reef tundra
Dont worry just keep driving
vagrant pagoda
I want to r2.8 swap a 2 door jeep
vagrant pagoda
Yeah
crisp granite
mhm
ngl, I want to get a larger car
but...parking would be a bitch
and I already have something fairly large
crisp granite
just drive a motorcycle, never worry about parking
I like keeping my body in one piece
not in shitloads of particles on the road
but, good idea
reef tundra
Get good protective gear atleast and not just your jeans
misty pier
back where I lived, i'd rather drive a motorcycle
crisp granite
that and, I'm kinda a spazzer
misty pier
way too dense for traffic there
crisp granite
even driving with my twitching and other fun quirks is a really bad idea
a motorcycle would make it way more dangerous for me
reef tundra
I mean, having uncontrollable movement on a balance vehicle would be unwise
reef tundra
Better before than after
crisp granite
my car has fairly loose tolerances, so a bit of twitching or something like that doesn't do anything
like, the steering wheel has a bit of play to it
reef tundra
Most steering comes from the power steering nowadays so it would be feasable to program larger tolerances
reef tundra
Russian engineering
crisp granite
so I got like 10 degrees of play for any twitching needs
crisp granite
Russian engineering
actually, yankee design, henco en mexico
misty pier
im glad i got a modern car since it doesn't do weird shit but
crisp granite
uh
misty pier
the electronics are the real pain
crisp granite
my car is fairly modern
misty pier
i have to take my car in for a recall fix
crisp granite
only 9 years old iirc
misty pier
mines was from 2017
reef tundra
Modern cars with automatic transmission could be hacked into to force it into neutral
misty pier
got a nice new car in high school 😎
crisp granite
cool
misty pier
it doesnt really scream anything special
its just a 4 cyl car with cvt
chugs through a whole week on one tank which is good but
gas is cheap here so
could be better
i still want an electric car someday but the range is what bothers me
umbral knot
do you think anybody has ever engine swapped a smart car
reef tundra
Engines are expensive
misty pier
is the engine as big as the car
crisp granite
chugs through a whole week on one tank which is good but
nice
mine can do like 2-3 weeks
reef tundra
https://youtu.be/2LPnRRvgOzQ
Would it even be street legal?
misty pier
ohh, i do somewhat long commutes each day before corona
so it got me through a week without needing a refill
reef tundra
To spite tall people?
umbral knot
I have a non street legal gokart that I drive on the street
reef tundra
Anything is legal if you dont get caught
misty pier
if i took a gokart onto the streets here, i would be flipped over in a ditch somewhere because of the potholes that riddle my entire road
its so bumpy
reef tundra
US INFRASTRUCTURE IS A JOKE
crisp granite
US INFRASTRUCTURE IS A JOKE
mhm
umbral knot
with my gokart I have to worry about the people ignoring the speed limit and going 90mph down a winding country road that people die on every day
umbral knot
with my gokart I have to worry about the people ignoring the speed limit and goi...
other than that it's quite fun
misty pier
i live on a hilly road and people here don't know how to drive
reef tundra
Remember speed is not what kills people, rapid deceleration is