#šātech
1 messages Ā· Page 1034 of 1
https://twitter.com/greymon55/status/1418454047888138241 ey, slight chance we get zen4 in Q3 and not Q4 of next year (very slight)
I can also confirm that AMD's early 5nm package validation has also started in China. https://t.co/lwVtgemXvR
Make sure you only use slots 1 and 2 when testing, @pale sigil
Wait what
According to the QVL your memory is only guaranteed to work in those two slots
But then it wouldn't boot
Itās done lads, the new lab.
Home Lab 2.0. :D
Wonderful in itās jank and awful design. š
Ew miner
:D
ewww, split monitors with a mic in the middle?
Planning on getting a G9 soon, but I needed more screen.
What GPUs did you steal from gamers
3070s. 17 of them. :P
Christ

Mate, I purchased them from gamers with the skill of haggling.
How do you even grt them
Haggling.
Nah.
In the UK at least
You just need to check second hand sites.
I mean hell, if I had a more stable rep here Iād offer to buy and ship them.
Lad
they are like £900-1000
Whatās that in AUD?
Something like 1500 to 1600?
1.8k aud
Lmfao
I mean, I have a dude that has 7 of the things to sell. I can see about shipping one if you want it.
Where ere you
I would've been able to get a 3090 and 3080 at stock prices yesterday
but the drop happend while I was doing a presentation 
Tragic.
Ha
Shipping them to the UK
Also tbf a similar thing happened
There was a 3070 drop for £459
Mad
Dude, the miners are so quiet now.

I don't even hear the constant REeeeeeeeeeEEEEEEeeEeeEEEEEE of the server PSUs.
@pale sigil Yo, gimmie a UK postcode so I can estimate shipping.
it was the first FE drop in like 3 or 4 weeks too, so would've been ez resell money 
no to scalping
Naughty
pfft I got 3 cards for MSRP price for friends
Quin, I'm a friend.
any more I'm entitled to get something ontop
"ez resell money"
I feel bad playing Minecraft on a 3080
well shipping/import would be a big hit in either cases
Implying I'd actually play New World. 
Eh, even with imports it's likely more reasonable than buying a 3090 here.
Trade me for my 1660 instead ;)
Dw, gonna do that to when I get 3080/ti 
get Bedrock and use them RTX shaders :P
Tbh kinda prefer seus ptgi
Why dont they make non RTX shaders so the plebs with non RTX cards can play with shaders too? ;-;
I'll sell you a Gigabyte Eagle 3080 if you want.
Non LHR.
Painful.
All rtx shaders aren't rtx
Why would Nvidia do that?
And can be run by anyone
On Bedrock you can only have shaders with the RTX cards
Issue is, I am looking for exact model of card. First was thinking of at least two, waterforce wb 3080 or ti, but since 4K and mods can eat a lot of vram, which can have huge impact if you consume too much vram, donāt wanna risk 3080
That's why nobody plays bedrock.
They are fucking terrible
its better for servers and things
Well continuum afaik had update where it now is.
YOU CAN PUT A WATERBLOCK ON ANY CARD, BROSKI.
._.
yo
dangerzone
š¤¦š¼āāļø
1660 can actually do RTZ
I mean, it is possible to have vanilla servers with Java/bedrock cross play
Just poo same with 1000 series cards
No, no it is not.
I mean, one thing itās better for is performance
Paying for every game mode is not better lana...
Bedrock can't have plugins which is where 99% of the general server stability and functionality comes from.
Yeah it is, you can run a fuick ton more players on less hardwaree than Java
But some bugs really suck. Also that *
No?
Yes
Thats why HiveMC killed its Java server,
Bedrock is more efficient
Where did you hear that?
They wanted $8 off me for sky islands or w.e its called
Bedrock is easier to scale, that's the benefit.
Unfortunate. Used to play on hive, sometimes quite a lot. Dropper, TIMV⦠it sucks they killed it.
And im pretty sure a cut off when a user buys access to a gsme mode or skins
Bedrock is wonderfully bad
Honestly, java is just so far better.
This.
I mean mostly, but Java has its issues
Like bedrock is amazing
The deathrun on Bedrock too easy ;-;
But really bad
I just want java mc to be remade into bedrock
Same freedom I msan
Imagine mods and freedom in bedrock
Then it would be Windows exclusive
No bs marketplace
on PC
I do not even care.
The people on other platforms can either get shafted or just use a VM or compatibility layer.
Sounds like a not me problem.
Especially the efficiency of bedrock with render distance
Ngl
It is a problem. There are quite a few mc players on macOS
^^
It blows me away playing bedrock at 60FPS
I did say compatibility layer.
On my phone
Would it really though? There are c++ games that can run on macOS afaik
And use the Android port on Linux
The game runs fine but the launcher is a piece of shit janky mess
If you get it to work, then the game launches fine
at higher render distance too!
change to game would make it not work on mac? oh no poor 5 people playing it on mac
Question is, since continuum has to already redo a lot of stuff to get rtx working, possibly even their own graphics engine. So maybe they could make more efficient shaders.
Actually there is a real number of mc players on macOS, especially with mc being such a widespread game.
@ashen spindle xD bexin wands a world
there is 9 people I know, and thats just me xd Im surte there are a fuck ton of Mac players
One thing I really like about bedrock is trees in snowy biomes having silver leafs
yeah probably same % that the rocket league userbase was, which was 0.3% on non windows platform
Nope, would say at least 5%. Definitely 1%
Minecraft is MUCH MUCH more popular than RL
ffs man xd
Of course, bedrock still only has a unchangeable 4 chunk sim distance
It sucks, big time
Nobody with more than two brain cells would run a bedrock server
This.
Bedrock is hell for any server admin.
No plugins, no mods, limited backend options, etc.
It's shitbox.
Because macOS users usually donāt play that many games, so there is a lot higher chance of them playing just more widespread games. Also mc being easier to run. And many people that donāt play other games much, which is especially on macOS play mc. And lana knowing 9 people is an example, that is not most likely just 0.3%
yes but that isn't something you can find unless mojan/ms would post about it publicly
no, lana knowing 9 people is the perfect example of cognitive bias/survivorship bias
So letās ask how many people lana knows who play rl on macOS
I'm making fun of macos not being the best for gaming and making something better for 99% at the cost of 1% is almost always the correct way
that's all
Well, nowadays with M1 for photo editing it sometimes makes more sense. But yeah, there are many people buying macs for dumb reasons, like just brand. But thatās especially true with iPhones
yes ideally you can use whatever, but I don't thing it's realistic nor the right way
Sure, but for games as big as mc
having thousands of dev hours wasted on supporting every extra linux kernel in existance over making new content
sounds like a bad idea
big game doesn't matter unless you can show me the % of players on mac is also big
1% is 1% regardless if you have 100 or 100million players
Sure, but itās lot easier to allocate devs to it when you have 100m. Even 1 million is a lot, they paying 20 million ā¬
^
yes but in this case supporting 1000 more people, also means you have 9 million instead of 99 people who can't get a better experience because you wanted to include everyone
Mobile is an abomination
Not really. With mc budget it does not really affect it much.
Tbh kinda surprised how long cliffs update takes since modders with lot smaller teams have made great biome modes in less
And caves existed for years already
I'm not even close to saying mac/linux should flat out not be supported, just if it is too much work it is absolutely the correct decision to cut it out as long as the playerbase on it is not massive, you're just rushing to defent apple again 
And also nether update
Give me a game like Minecraft that looks nothing like Minecraft and I'd be semi-content
All existed years before via mods
I despise the cubes, personally
Thing is, mc is not 1000 times as hard to get to macOS as to game with 1000 times less players. So itās āeasierā for the devs to get it to macOS compared to small games.
what
If it came at the cost of getting 300 fps I'd care less, I think, but for the performance you get, ugh
Valheim, unironically
I have no clue what you tried to say
Space Engineers doesn't do it well tbh. Most games don't. Valheim does it decently I guess
We're still a long ways off to making my perfect survival game though I'm sure
Doesn't MC work natively on MacOS anyway, because java?
Why? Java is objectively superior
objectively superior in what way though
If it was moved to another language but wasnt locked down to only being able to buy stuff from the store like Bedrock is, that would be the dream
Bedrock still hasn't reached feature parity
Some features behave differently also
This means that you have more profit, and developing for macOS will most likely make you money with game like mc, which you can then actually use to improve game more. Just gotta have more devs
or have the same people work on making new things that the other 99% can enjoy, which will cause more sales than having the other 1% be able to aplay most likely
I don't think porting to MacOC makes sense if you can't expect at least 10.000 sales on the platform, probably more
Not if you hire new devs to develop for macOS. Which in case of mc would very likely pay itself, since 1 million is a lot
yeah or you hire new devs to work on another set of new features instead
same thing
Unles the 1% of players is the 1% š
True. If you arenāt making universal app from start at least. If you can get that many sales however.
Nope, improving game like mc most likely wonāt make you as much money as porting it to macOS. And again, you could do that, however, you can also hire devs to port it to macOS, with so many players making profit, that you could use to hire even more devs for base of the game to improve, that you otherwise could not
you can repeat it over and over again and it wont change
unless you can provide a solid number that shows how much it would make, it's just wild speculation
and any estimates that one can make from other games in similar situtation points to the exact opposite
that doesn't mean it would be the exact same, but it means there isn't any point to stand on to argue otherwise
The Freedom Phone by Erik Finman is an elaborate waste of money, and quite possibly also a security risk, do NOT buy it! Try Dashlane for free on your first device head to https://www.dashlane.com/boss and if you decide to upgrade to premium use my code boss for 50% off!
Subscribe for more content (IT'S FREE) : https://goo.gl/pLg6fE
It would ma...
The marketing video alone told enough that it is a scam.
Navi31ćÆ6x10ć«ćŖććØē¢ŗäæ”ćć¦ćć¾ćććć©ć®ććć«ę§ēÆććć¦ććć®ććććć¾ććć
Navi33ć«ć¤ćć¦ćÆćä»å¹“ć®ē¬¬4ååęć«ćć¼ćć¢ć¦ććć
15k shader cores
I'd be doubtful, but after Threadripper... you know, if they can do it, they will
Gimmedat
mc java can run faster than bedrock with optifine and speed shaders
dafuq are speed shaders?
The fuck is a speed shader? 
Presumably higher performance shaders, but you don't really use shaders for that.
That's mostly up to mods like Lithium, Phosphor, FoamFix, and Optifine.
Never underestimate the craziness that is Phosphor.
Or Lithium, for that matter.
the LiNa + Phosphor is nuts on something like a pi
Interesting
is that just because of cache hitS?
Past the first level cache, RDNA 2ās L1 and L2 offer lower latency than Ampereās L2. Infinity Cache isnāt as low latency as we first thought. It looks to have around 156 ns latency, making it about 15 ns farther away than Ampereās L2. It still does quite well for such a large cache, and should provide benefits beyond preventing a VRAM bandwidth bottleneck. Requests served at lower latency from Infinity Cache will mean RDNA 2 doesnāt have to hide as much latency.
i would like to see the same chart
but 6800 vs 6900
ive always wondered if have more cores with the same memory configuration bottlenecks
That's using different tests, though
Interesting.
they mean strix?
oh
they are talking about the load
so that isnt even apples to apples
Nope, literally, their old vs their new testing pattern :p
Random is more accurate
yeah what dose it look like with the same

wait u wanna run cemu on streamdeck?
daily switch shilling time with locke
Where'd you dig that up
His friend made it
It's for playing uh
Any platformer
2d or 3d
Fighting games
Streaming from your PC across the house
š¤·āāļø mediatek :/
But why not 2d and 3d platformers?
That's up to valve
The Internet 
@zenith sonnet ;G_G;
Here the link @fickle ore
Manjaro only got gnome 40 yesterday :O
aha
Reinforces my believe Manjaro is budget-arch.
Arch but generally crappier with more packages preinstalled.
Don't get me wrong, I really don't mind distro's that come with a gui but when you deffer updates to make it less compatible / stable...
They have an odd mix of upgrading core packages and then that causes issues when they hold back software that uses those packages
Nvidia and amd being lower really surprised me tbh.
And you'd be 100% right about this
Yeah, donāt remember the last time mine fell out of my ear. Especially after I have gotten used to them.
Samsung gave me for some included in the price of my phone and tbh they are very very convenient
Also always used to destroy cable of wired ones.
And now also almost use only one AirPod at a time. Apart from times noise canceling is nice to have.
I've never lost mine xd
Although I just use them in bed
Even the new ones end up broken
Lightning is a horrifically fragile connector...
Managaed to destroy couple of lightning cables too. Especially when charging while lying. Lightning itself was not an issue, but the cable
So many broken cables in ewaste
But MagSafe solved this for me
Only about half of the ewasted cables are cable fraying
Wish you got the same cable with phone that you get with HomePod mini
(Braided cable)
Oof
Yeah, but rarely. One you canāt use with other devices, well, two, also iMac one, and then the TB one is extremely expensive
I need good earbuds that I don't have to constantly push back into my ears do any of those exist
Yeah....
The tb cable is the worst tb cable
Every earbud I've tried fit fine
Other than the ones with no removable ear tips
Like airbuds
Try the foams
But another nice thing about wireleless earbuds is not needing a phone, and being able to just play songs from a watch
I mean the braided one
Bet they only work on apple watch
They don't sell standalone a TB cable
That is braided
It's the destructible garbage
They are Bluetooth so with any Bluetooth watch afaik
Yup
Probably more reliable because all wired things I've used break in the exact same way with the cable
Oh shit
That ain't being sent out to retail
I use magnetic cables
And uh
They don't work with apple
Yes the lightning ones
No they don't work
That doesn't apply to the NFC on iphones but ok
Well, not sure, but since you can connect AirPods to android and windows devices, donāt see a reason you could not to some non Apple Watch
I guess so
Spatial stereo is pretty Pog. Not that I used it a lot since I usually use just one airpod at a time, same as rn.
Galaxy buds non pro or plus for sound quality
Yes
Nowadays many can be to some extend standalone, some even supporting eSim.
Er2xr are good but tonality is meh
Arias are nice but painful in some music
I haven't seen any recently but the mh755 are god tier
Wish Apple Watch could do spatial audio, or Dolby atmos.
Chi-fi?
the homepod braiding š©
i love it so much
also, @astral haven, don't cheap out and get the 0.8m one
you'll regret it
Itās Pog same as itās sound. Donāt use my HomePod much though, since I mostly use AirPods now.
comparing a speaker and 2 teeny earbuds
tf bro
Just saying why I donāt use it that much anymore, not comparing.
btw
they added "reduce bass" option to homepod
or something along the lines of that
literally night mode
lol
If I get a thunderbolt cable it's gonna be 50m now
100m, come on. dont stop there already
50m is the longest available
there are fiber ones
50m is the longest available
E X P A N D
40gbps
Wireless wires
š¤Ø
cablet
Scam. lets sue apple now. court will be fun
Apple
@digital cliff Were you still interested in a Modi 3+?
Ok
I wonder though
Why no amps?
Can't charge over a fiber cable lol
With strong enough lasers and efficient enough receivers
Or just a copper cable along side it
Then it's not a fiber cable only...
No, no you cannot...
Which is pretty much an led
Just need the sun to be 10000x brighter for that to be enough
i didn't say it's a good way to charge
Laser would need special shielding to not blind anyone looking at it. 5mW is the threshold for legal limits as eye safe.
The fiber cable is just to guide the laser and keep it from attenuating too much
it's existed for years
Yeah, then it definitely has issues. Rain sounds not the best, live listen makes everything way too louderā¦
Jailbroken?
Nope
The UI of this is indeed strange
yeah....
Literally on the newest device of this type 
yeah, hardware aint fixing a UI design xd
Always blame the user -Apple
Literally
Their motto
"oh no, you just holding the phone wrong. Thats why the reception sucks"
Next phone release, they moved the location of the antenna bands
so...how do I get my hands on a Ryzen 7 5700G?
any word on a proper release soon?
anything at all?
Quin...
I have my reasons
it's more that I want an 8 core...and I'd like an iGPU in case my actual GPU goes down
remember it's slightly worse than regualr zen3 in cpu power
ĀÆ_(ć)_/ĀÆ
it's more the 8 physical cores that matter for me
it's not the end of the world if it's a little weaker than a 5800X
it has half the cache which is where a lot of the gaming performance comes from
idk, depends on game, resolution, settings etc.
think gn did a review of it maybe already?
that's why the extra cache on zen3d will be relevant š
Q4 this year for those
hopefully I don't need a new, pricy mobo for it
noice
AM5 starts with zen4
since alder lake is coming Q3/4
and zen 4 in Q4 of 2022
the 3d stacking cache is amds bet on being enough to keep a lead
alder lake is gonna require a new socket, no?
that is quite cool, but... at least to me, pointless
no devices out there use DDR5 or PCI-E 5...yet
new GPUs are PCI-E 4, and I don't know where to source DDR5 ram sticks
I more meant to say that to have a point that zen3d is gonna make a big push despite "only" changing the cache size š
well ddr5 is in volume production already, just not much sense to put it on the front of newegg or w/e big consumer store when consumer boards are not using it yet
as long as that gets dumped on the market along with LGA 1700 mobos and CPUs supporting it
one of my buddies wants to upgrade from tenth gen intel to alder lake when it releases
...
yeah
I don't get it either
hey at least it's an improvement
it is, but jesus christ, value for money won't be there
when you've been here when bexin decided to wait for rocket lake when zen3 came out just so it can decrease performance 
he'll need a new mobo and CPU at least
oh amazon seems to have first kits just like newegg already https://videocardz.com/newz/teamgroup-elite-ddr5-4800-32gb-memory-goes-on-sale-on-amazon
TeamGroup DDR5-4800 memory now available for preorder, shipping soon The first DDR5 memory is now available for sale. TeamGroupās DDR5 memory, which only launched this month, is now available for preorder on Amazon. A 32 GB kit with JEDEC DDR5-4800 specs is listed at around 310 USD. This is the first commercially available kit before [ā¦]
The graphic makes it feel like a fake chinese seller
welp, I'll kinda bad for him now....
ram alone will be $300-400 CAD at least, since he wants a 32 gb kit
I mean thats the first kit, we dont really know how prices will be
true, but I doubt they'll be current DDR4 levels
- there is a good chance that we have something like skylake did and it actually has a controller for ddr4 or ddr5 and you can still use either
that's what I was thinking
so as long as you can get him to hold off till reviews and prevent him from preordering, no need to feel bad afterwards

it's a pretty decent kit tbh
works too I guess
and it's a 2x16, so would be good for my inevitable jump to ITX
I could save 50 bucks there, use it for some overpriced white-boy coffee at a starbucks, or use it to buy a new video game
ĀÆ_(ć)_/ĀÆ
we need more dtx
most itx rigs have 2 slot full length gpus anyway
so a 2 slot full length dtx would be a good match
4 dimms
2 m.2
good audio and vrm
but still fits most of the same cases
I think with next gen we will need more bigger boards and not go down in size 
bc power draw go brr š
nah, going based on what leakers (with good track record) hinted at, it's gonna be ore watts on greend and red
for gpus or cpu?
gpus
because size can stay the same for gpus
Quin...I understand that
we have KW sfx psus
yeah but more power = bigger coolers
also...the novelty of a PC the size of a console that absolutely shreds
would a 280mm rad work?
i have prossibly the worst case made for my vr rig
modded to hell
only because of the size
idk, with chiplets I don't even know if the same cooler design would work of if they need to change parts
well depending on how they are spread they might have to adjust how heat pipes and such have to be laid out, and high end chiplets might still be a decent size bigger, dont think we know that yet
or maybe im just overthinking it, won't deny that possibility haha
idk, the rad I plan on using has a gigantic square block, I'm sure it'll hold up
well with waterblocks it shouldn't be an issue, those aren't gonna be tripple slot or extra long lmao
I'm sure it won't be quad slot next gen
because that would be completely nuts
nah I don't think so either because that would be causing issue with sagging and all that
was more thinking of increased length
EVGA 2080tis had a fantastic bracket
Pushing down on the back of the card it still felt relatively rigid
Also it could be a thicker heatsink with lower fin density
i mean there are 4 slotters now
Yep
length causes sag
There's 1 slot up 3 slots down for example too
It's just weight
Weight is sag
sure the weight is more evenly distributed, but more metal surely still pulls it down
Weight without reinforcement
Ye but it's external
like you put it on your psu shroud and it holds up the card lol
also its 4 slot
3.5 down and 0.5 up
This is the 2080ti bracket
the 3090 aorus is like 4+ slots isnt it
ah there we go
"the power consumption of the navi31 will not be as low as 300W.Gesicht mit FreudentrƤnen " š
..thanks clipboard for turning an emoji into the description
and ada supposedly the higher powerdraw again
the 6900xt is 300W, 7900xt will be higher and 4080 even higher

as long as it's not a 4+ slot design longer than 330mm, I'm good
and I hate the 2.x slot bullshit
if it's not a 2 slot card, it's a 3 slot

the 2.5 is so it can breath if there is a card next to it
if you put a 2 slot card up to another card
you get shit for air
I hate marketing it as a 2.x slot
also if you have a small case that only has 2 pcie slots
it's either a 2 slot, because it matches that standard
then its right up ot the side
or it's above that and a 3 slot card in practice
if we're lucky chiplet means we can cool it easier/in the same way as current powerdraws because the headspots are spread out
could honestly go either way
Hahahahaha
Lmao
Tsmc liquid cooling on die will become required quickly
That's normal 2 slot...
Yes?
It's almost as if making a competent dual axial is tough to fit in 2 slots standard height...
I really hope elmor can make some flexible nvlink bridges lol
why
Cause I wanna support elmor and not Nvidia
And I also want to nvlink
but why flex
that reminds me, with next gen platforms, wouldn't CXL just be a better way than sli/nvlink?
Cxl doesn't mgpu
yes it dose
And nvlink is faster than cxl 2
Shared memory is alot easier than shared compute
Cxl is a replacement for pcie that supports 2 way dma
cxl is based on pcie5
Yes
which means is on par with nvlink
256gbit vs 800gbit?
desktop gpus only have 4 lane nvlink
Nvlink is 200 gig per lane
4 lanes would be 400 gig equivalent
its 50
For turing
ampere is nvlink 3.0
which is 50gbit a lane
desktop gpus only get 4 lanes
titanv and quadros/teslas get 6 lanes
the socketable accelerators have 12
i think the a100 pcie have 12 lanes to
they take 3 bridges
š¤·āāļø the a100 does 128GB/s
Oh
has someone found out how to sli 3090s or are they only compute?
Nvidia showed a demo where they were moving 128GB/s through a gpu
so its only 300Gbyte duplex
Pcie is in Gbps...
yes
that is a100
not a consumer gpu
it dosent even have video outputas
desktop cards have 4 lanes
4 lanes at 50Gbit
Desktop cards have 1/3 the a100...
no they have 4 lanes
Nvlink is horriyingly fast
Cxl is fastish
But here's the other thing
Cxl overhead is smaller than pcie but is still a thing
And nvlink has zero overhead for direct links
And only ads on the order of 10ns going through a switch
Even if they were the same bandwidth that latency and overhead stops it from being a direct replacement
And here's another thing
Nvlink can be used alongside cxl
2080ti nvlink was 100GB/s already...
NVLink is enabled in different ways depending on what video cards you have, so we have compiled instructions for multiple GeForce and Quadro models. Once you've gone through those steps, though, it can be helpful to have a quick way to test and make sure it is working. We have put together a small program that does just that - and reports which ...
Sli was 2GB/s which was really not enough
I just want NVLink back in consumer shit.
I wanna do 4 way GPU scaling again, damnit.
3 or 4 way nvlink would be very expensive
But the thing is, sli scaling is no longer possible, and ampere is proof that wider is not nescessarily better
I know, I know.
eh
Just always liked SLI and NV. :(
Sli isn't work for devs
and dont have sli
SLI is engine work, iirc?
no
...
Explicit mgpu is in engine
That's not sliā¢ļø
True sli was implemented by Nvidia and you just had to chose which of the many many profiles worked best for you
Sounds to me like integrating it into the engine is a less stupid option.
it uses nvlink connector
If it's a profile the devs have to do it.
Explicit mgpu uses zero Nvidia ip
Therefore it is not sliā¢ļø
If it's engine they're all doing it by account of using the engine.
That's incredibly tough
Tough work once?
explicit mgpu is really fucked, it's nothing like sli
Nope, it's tough for the devs too
also sli kills 1% lows
They have to dispatch work to 2 devices
which is why its dieing
On some profiles
Nope
What's the 1% on NV versus SLI?
source
nv link is just the connector
Source being I tested it
There are sli modes that are alot less stuttery that just aren't as good scaling
everyone else
Iirc the bandwidth is higher, no? If used right it'd be not as bad?
has the opposite result
Noone tries actually tuning sli...
There are about 60 different sli profiles and testers always use the default one...
Devs usually chose the highest fps one or the first one that works
Same at lower resolutions, but the sli bridge dies at higher resolutions
Ah.
Well
Same relative to averages
It's alot higher because the GPUs are alot faster
Untuned sli has identical min frametimes to sli off but with bad jitter
Tuned sli either has same frametimes without jitter or minor jitter with properly scaling frametimes
Yup, you just enable it in Nvidia inspector
Usually poorly
Put of all the profiles only 1 or two are of reasonable performance and often never over 1.2x
And without exception beyond 2 way is negative scaling in my experience (For games without stock profiles)
Tho I have gtx 690s at 1080p so new stuff and higher resolutions may work better
Why not?
1080Tis get capped at 1440p.
In games like Destiny they're actively bound.
Are they also below 60 fps?
With decent settings? Yeah, sometimes.
Well then that might scale
Pushing everything into ultra 1440p can dip them down into the 60 range in some games.
Or things that're poorly optimized.
š¤·āāļø if you have the cards might as well buy the bridge
On it also hurts sli scaling if you are reducing bandwidth to the cards, so you generally need hedt...
Ryzen 3900X do?
1080ti isn't
Unless you've got a Broadcom gen 4 switch x8 per GPU is what you get
Also the 3900x isn't a particularly fast CPU
It's great bandwidth but not super fast
Mmmmm.
Depresso.
Read and write are relative to the CPU btw
So yes that is a 9900k with 512gbps of effective pcie bandwidth
4.0x16 is 256gbps
And the 9900k is 3.0x16
not true
just never production
[citation needed]
z490
There aren't any gen 4 switches under $250 bulk
the asus z490 prototype had plx
Only 3.0
you said x16+x16
im talking amd z490
before intel did z490
amd had planed a updated x470
called z490
Called x490 you mean?
I've heard of x490 but z490 is gonna [citation needed]
Yeah, gonna have to agree with [citation needed]...
[citation that isn't a picture needed]
citation provided
[preferably a spec sheet]
AMD is reportedly readying a new high-end motherboard chipset positioned higher than even its upcoming X470 chipset, which makes its market debut on April 19. The new Z490 chipset, as it's called, is emerging to be significantly different from X470, and could finally overcome its biggest limitation ...
i cant find the asus one
but asus had one with a plx
Nah that's fine
That's x16 +x8 tho
š¤·āāļø plx ain't for multigpu
Plx is for pcie aics in general where you need a ton of them
See the x299 sage
odd considering that boardcom mentions gpus in every pex document
GPUs
Not motherboards
Corroboration is not causation, friend.
Plx on gpu is great
Idk how this is confusing.
Plx on motherboards is janky

But for normal aics it's fantastic
i mean
See the x299 sage
i own 3 motherboard with pex chips
never had issues with multigpu on them
they do add some lag
Yes
but
They don't add lag for storage applications or networking or audio...
Not if you configure them properly
i mean they still do
100ns is best case
source have one of thier 10g cards that uses plx to allow 4 ports
also i have tried to use one for nvme
Ah but that is fixed
So you can compensate for that
And for networking you have larger buffers than that anyways
So you are better off taking the 100ns transfer start penalty to get the transfer done 3-4 seconds faster
yes it better than lower bandwidth if you are saturating the cpu link
but highend nics do routing internally
Not from storage they dont
my intel 10g will route though direct on the buffer
The buffer is just on the card...
the avago wont
as all 4 ports are separate devices technically
and the only reason to use a pex in a nic is for routing
as the total thoughput isnt going up
I'm talking about plx motherboards
its the same
There are a million and one reasons to use plx on aics
now gpus did fine
Its absolutely not the same
because most gpus run fine on x8
so the dual gpu cards didnt care
but 10g was hell on 2.0
They run fine
But they run worse than hedt would
With storage, networking, and audio, you want the faster CPUs that are on msdt (except for zfs and similar where you need coars)
And that's where plx helps you out
You have these pcie accelerator cards that are not all being simultaneously used, so plx does just fine
š¤¦
40 gig Ethernet can be entirely saturated by a single x16 link shared between the data source and the nic
I mean my server is encrypting and decrypting the data
Data comes from the 3 2.0x8 devices and flows out the 3.0x8 nic
i love the baby noctuas
i have one of those super tiny tower coolers on a pi
with them on it
Tbh, why go passive when Noctua ULN exists?
They never break, cool pretty well, make no noise whatsoever, etc.
because its not actually passive
I know, but I'm more talking generally.
You see people going to absurd lengths for passive when a ULN makes the same noise, really.
Why?
because thats actually not true
there are some people who will actually properly treat their theater rooms, and a fan in general will make some noise
Wait...
If you're treating a room like that why do you even have a PC in there?
That's already quite the amount of work to go through, one could just Thunderbolt 4.
And then it's proper dead silent.
why tb4 and take up room in another room when you can have it contained in 1 room
and a passive cooler is alot cheaper than a tb4 cable that will go between rooms
what
Sure, but then everything isn't in a beautiful server rack. :P
The clean aesthetic is great.
š¤·āāļø says you
you can have a mini server rack with all your audio gear and your pc right under your tv in your theater room
you can mostly enclose a passive system
But then it's not in a single 48RU. :P
Because the rest of my shit is also in the 48RU, and I'm not having that thing in an audiophile room.
Hence Thunderbolt.
And then what's the point?
still silent and now you have the knobs to your amps right behind you
No way I'm putting my servers and such in a 48RU in a theater room, especially since it'd do double duty as an audiophile room.
Nah, I'd run thunderbolt for the interfacing stuff (screen, kb/m, etc), and have the rest of it in a dedicated server room.
Like... I could buy a Roon Nucleus, but why would I do that when I could spend a fraction and make something better in a rack?
As far as the amp stack? That'd likely stay in the audio room for a couple of reasons.
Unless I could pull some complex shenanigans, but that's too much work.
you ever watched z reviews?
A few times, but only as a curiosity.
Hey man, what works for some people works for some people.
My equipment is really airflow heavy, it's noisy with so many fans.
1 here and 2 more on the other side with computer stuffs
yea
but audio rack equipment is silent
Not all of it.
There's a few things that do require some fans.
But that's fairly simple to fix.
ok
studio rack equipment is silent
tldr is passive pc has a use or two
thats not many but thats enough to exist
I wanted a nhp1 because i believe its the only cooler that it would be safe to ziptie a delta to
and i wanted to see
tisis core + 3 arctic fans or nhp1 with 1 delta
yup
there are 2 things low fpi is good for
passive
and HIGH airflow
im also curious what the minimum speed on the deltas are
cause the ippc can go down to barely over 100 rpm
where more normal fans can only go down to the 300 range
if the 5600 rpm deltas can go down into the 100 range that would be ultra pog
im confused
Why do people argue so much about which OS is better?
Itās so simple 
iOS 15 vs Android 12?
15>12
iOS>Android
macOS 12 vs Windows 11?
12>11
macOS>Windows
Simple 
get a webserver that you just use as storage place or also put plex or w/e works with your data on it for 5$ or so a month
google drive though I kinda feel like you'd not want google
netcup is what I use, linode seems to also be a decent option
What is f51ae?
oh and just as confirmation, it's an evga issue that cards turn off with errors https://www.igorslab.de/en/evga-geforce-rtx-3080-rtx-3090-and-not-only-new-world-when-the-graphics-card-goes-amok-because-of-design-failures/
any other cards just have regular thermal protection kicking in due to 4 digit fps
Except that a tripped protection circuit on a Radeon card is logically something completely different than a blown controller chip on a botched board design. The protection circuit is there exactly for such a thing, only that with the fan one must question evenly.
Not really a fan of this distinction. Dead card is dead card, Igor...
but amd cards don't die..
it's purely the evga cards
though I do admit the english versions of his article are a bit unclear sometimes, it's not his strenght to translate it well š
ye
I was just saying... in that example, the Radeon card isn't dead, only needs a reboot, right?
Seems like, yes
Another thing is with a decent number of players you expect at least some users to have random problems.
CTR is weird
I got Silver Sample (with everything open and running), then Gold Sample, then a random reboot, then Silver again
Anyone got some ideas on 3D printers?
Looking to spend maybe 300 AUD on it, so far the classic choice of the Ender 3 Pro is winning.
Just a normal PLA/ABS/etc printer, not resin yet.
Have printers become significantly more expensive recently?
I dunno, I remember looking at them a while ago, and now they seem to be twice the price if available at all
Looks like printers went up in price...
Prices go up 
Günstige Preise, Meinungen und Bewertungen zu vielen Produkten findst du bei heise online


