Good
#πβtech
pale sigil
shadow minnow
Someone saying 3080 is 900$ (person in NVIDIA discord, saying he has been trying to find links... on nvidia website)
ashen spindle
Price leaks mean nothing
pale sigil
Eh
ashen spindle
No, nothing
ashen spindle
oh yeah lol
shadow minnow
I know, just saying
pale sigil
Just fuelling their bs
ashen spindle
Someone is saying 3080 is 1276$ and releases at 7:41pm during a bloodmoon
ashen spindle
some of these will be correct too, not too many numbers between 800 and 2000 that would make a round msrp after all lmao
crimson wigeon
Meh, people are gonna say the 3090 is way too expensive and it'll still be immediately sold out so whatever
ashen spindle
tbh that card doesnt really matter, they can overcharge there for all I care as long as the others are fine
since the 90 is basically a titan marketing for gamers now
vagrant marsh
any1 here who is dutch who knows anything about memory?
vagrant marsh
33 minutes!
crimson wigeon
Ohboyohboy
vagrant marsh
happy happy joy joy
right
wont stop an afghani with an igla to shoot them out of the sky tho π
ashen spindle
speaking of future... is intel waving the white flag? no longer ignoring zen optimizations?... https://danieldk.eu/Posts/2020-08-31-MKL-Zen.html
vagrant marsh
uhh
Once we start talking about a few million worth of helicopter, the soldier becomes more expendable
Hmm
True
I watch too many russian dashcam vids. You cant tell me they learned anything besides press gas to start XD
Try british or australian dash cams. The word muppet gets abused there XD
ashen spindle
kinda feel like taking a nap, I bet they spend the first 15-30 minutes about the history and how great nvidia was at pushing stuff
ye.. I was there in person
vagrant marsh
no cause i fell asleep
ashen spindle
did get a chance to talk to jensen for a moment so that was nice
I mean it was like 30 minutes away from me lol
vagrant marsh
You did that voluntarily or more for your job?
ashen spindle
oh right forgot he left, where is he nowadays?
vagrant marsh
My brother once got send to america for his job to fetch a new iphone at the time cause it released 3 days earlier in america than the netherlands. His company demanded they would be ahead of time in making their app compatible with the latest OS
ashen spindle
oof he went to intel
vagrant marsh
Whos tap?
crimson wigeon
Oh yeah what happened to all those pretty CGI ladies for showing the power of graphics cards
crimson wigeon
YEAH RTX ON
crimson wigeon
do the floss haHAA
pale sigil
55 MINUTESSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS
crimson wigeon
Until..
vagrant marsh
nvidia reflex making me competent again!!!
woohoo! this will be the age of the old people!!
seems like nvidia is really pushing AI aid
vagrant marsh
they said it was a super computer doing it somewhere
vagrant marsh
ahh
vagrant marsh
nvidia
:p
reinvented computer graphics...
say that to 60% of steam that has no rtx card
old marlin
40% is a lot more than expected
also real time raytracing is pretty amazing ngl
even the current implementation
ashen spindle
tech has always been amazing, marketing just twists it and that leads to missed expectations
pale sigil
45 MINUTES
old marlin
lol
crimson wigeon
Ah yes need AI to solve the awfulness of TAA
old marlin
I wonder if we will see low quality pathtracing with denoising
well you can use a lot less power to get 80% of the result
ashen spindle
doubt it, dont think the hardware is good enough yet
old marlin
it should be
denoising is a lot lighter workload
and can run on RT stuff
just split the workload in half
ashen spindle
I mean I guess it depends how far you take pathtracing
vagrant marsh
wait
old marlin
and you can keep frametimes high
vagrant marsh
what did he sa y about xbox
old marlin
by just not denoising too much
ashen spindle
ok, let me rephrase, I dont think pathtracing in a useable case is doable yet
old marlin
well yea
crimson wigeon
DOUBLE
old marlin
I still meant it as support for effects
will be insane workstation cards lol
ladies and gentlemen
we got him
vagrant marsh
samsung
ashen spindle
thats been known for a while now
old marlin
was it known as a fact?
vagrant marsh
made in korea
old marlin
ashen spindle
it was known tsmc did not have any capacity for nvidia, so kinda yes
crimson wigeon
The graphics are INSANE
zenith sonnet
@old marlin That graph start at 0.9?
old marlin
which seems realistic
ashen spindle
ok this looks wayy better than the demo earlier this year, im impressed
vagrant marsh
enjoy this marbles vid at 1440p. video quality doesnt go beyond 1080p
how do i enjoy the 1440p!!
zenith sonnet
1? So it went from 3 to 5?
crimson wigeon
Can definitely see some artifacting though, FAKE
ashen spindle
... no artefacts would be an obvious fake lmao
old marlin
gow fucking narzistic can you get
ashen spindle
crimson wigeon
Stop making games so big
old marlin
people are just not used to buying big hard drives
crimson wigeon
rip consoles huh
vagrant marsh
wait
old marlin
the last 5 year had barely a difference in the HDD space of average consumers
vagrant marsh
instant load with rtx?
old marlin
even though prices fropped a lot
vagrant marsh
INSTANT LOAD WITH RTX?!?!?!?
old marlin
no
vagrant marsh
he said that!
vagrant marsh
it doesnt go through system memory but through graphics ram
thats instant loading
same as ps5
old marlin
"our new fast memory can load stuff fast from new fast ssds"
vagrant marsh
i got 2 2tb nvme's
vagrant marsh
oh yes yes yes
old marlin
we got him
vagrant marsh
hes showing the new already leaked out design!!
old marlin
there it is
vagrant marsh
COME ON BABY
old marlin
lol
vagrant marsh
SHOW US THE MONEY
old marlin
the sounds 
vagrant marsh
double blower style!!!!!!
old marlin
not blower though 
vagrant marsh
ITS DOUBLE BLOWER
old marlin
it is not a blower
vagrant marsh
ashen spindle
yes
vagrant marsh
not samsung for the memory too?
ashen spindle
only micron makes this memory so far
old marlin
"straight out"
crimson wigeon
Ah it's gonna cook mah CPU
old marlin
rip CPUs
crimson wigeon
Waow dat price
vagrant marsh
september 17
crimson wigeon
Waow they didn't hike up the price!!
ashen spindle
1080 was 499 msrp btw...
old marlin
cute
vagrant marsh
and 3070
no1 expected them to show that one as well
they mostly expected the 3080 and 3090
but the 3070 as well
starting at 500,- in october
crimson wigeon
FPS ON
vagrant marsh
FPS ON
lets hope amd shows a response soon
they dont have dlss
and no matter how much dlss sounds like a gimmic, it truly works
old marlin
yea both even have guards on the impeller
crimson wigeon
Oh shit here it comes
vagrant marsh
yea, but those games do show it works
old marlin
to prevent radial air leaks
vagrant marsh
rtx titan
old marlin
lol
crimson wigeon
Damn that thing's chonky
vagrant marsh
GIANT GPU
GIANT AMPERE
3090!!!!!!!
THE MAGICAL 3 SLOT GPU!!!!!
thats a fucking big boi
crimson wigeon
Now how am I gonna link up 3 of them
vagrant marsh
they invited 4 streamers. none of them look familiar to me
why not invite pewdipie?
crimson wigeon
yeah literally who
old marlin
they also all played on controller 
ashen spindle
idk man, I wonder why they did not do that during a pandemic
vagrant marsh
wait...
old marlin
can't wait for 2k 3090Ti
vagrant marsh
why isnt he saying the price or release for the 3090?
crimson wigeon
But he did?
vagrant marsh
no, 3080 september. 3070 october. he only showed the 3090
ashen spindle
3090 september too
crimson wigeon
Wasn't it very soon and 1499 bux
ashen spindle
sept 24
vagrant marsh
still, the thing that exited me the most is that thing about instant loading through rtx graphics cards
thats the crap ps5 was boasting about
ashen spindle
ehh, close
vagrant marsh
no. not close. they outboasted sony
hey sony, look. we can do the same.
FUCK YOU DOLPHIN
and a fukka you, whale!
wheat plume
Can Firefox use the MacOS global menu? or is it like on Plasma desktop and cant use it?
I just see the app name
crimson wigeon
But now.. to wait for what Mark Bench has to say, if there's any truth to the marketing
ashen spindle
on a 1-10 I'd rate this release a 7.5 so far, now to wait for reviews
also AMD pls gibe navi
yeah reflex and IO look cool but needs some more details
pale sigil
i missed the stream
vagrant marsh
they talked about instant file loading through rtx cards with ssd's callum
they generalized sony's 'instant file loading' gimmic
pale sigil
Wait did i miss something
vagrant marsh
YOU MISSED HTE FUCKING STREAM
shadow minnow
When will ampere appear on nvidia webpage?
pale sigil
did they reveal the gpus
vagrant marsh
yes
ashen spindle
crimson wigeon
Yes you licherally missed everything
shadow minnow
3070 - 499
3080 - 699
3090 - 1500
shadow minnow
3090 new titan, 24gb GDDR6X
vagrant marsh
FOR THE HOLY EMPEROR!
ashen spindle
wait for amd....
ashen spindle
yeah on mobile cpus
ehhh not really, bitcoin no longer makes sense to mine on a GPU
needs special asics for it
pale sigil
sigh
ashen spindle
? I doubt anyone semi smart person still mines bitcoin
pale sigil
3080 gonna kill me
ashen spindle
and not something else if you use a gpu
pale sigil
Im gonna wait
vagrant marsh
now we wait for amd's response. can they top nvidia's promise of instant game loading?
ashen spindle
before they top it, nvidia needs to show that it works first lol
crimson wigeon
so this month
pale sigil
xD
nova carbon
so excite 
fickle ore
ehh waiting for benches before making decision.. company slides are company slides
ashen spindle
also, notice how he didnt even touch the power aspect? how much power the card draws
vagrant marsh
350 WHAT?
shadow minnow
They did in one slide
vagrant marsh
lets see if any retailers already put it online
shadow minnow
They said it is 2. As power efficient as Turing, but also goes up to higher power
fickle ore
those perf figures are probably for raytracing 
ashen spindle
yes aka he didnt mention it
pale sigil
Gaming is gonna be expensive to run at this rate
No playing games like BDO for me anymore
ashen spindle
so what was the thing that you mentioned earlier that released 5 minutes ago callum?
pale sigil
i dont know
pale sigil
?
crimson wigeon
You were counting down to something you flid
pale sigil
Oh, forget about that.
ashen spindle
shadow minnow
When will we see nvidia website update with ampere stuff?
ashen spindle
didnt locke already link to the nvidia site?
pale sigil
the blog
shadow minnow
Nothing else appeared
ashen spindle
tbh the most important part is, if we will get 3080 versions of AiB's with 20 instead of 10gb
bc that would make much more sense to get
vagrant marsh
the instant file loading. the ps5 works on amd hardware right? so...maybe amd already has it, and they will simply say the same thing? hat they will support instant file loading on radeons?
pale sigil
ashen spindle
oh yeah, no 12 pin mentioned?
also, keep in mind amd has been the first to intodruce lower latency stuff, image sharpening etc., nvidia had to catch up, so I doubt amd has nothing but hardware to show
ashen spindle
not like most gamers watch a GTC talk
pale sigil
8K with ray tracing.....
ashen spindle
with dlss
also, the amd team has moved π https://twitter.com/sherkelman/status/1300842481886662658
Twitter
vagrant marsh
they shown a sign of life?
pale sigil
i still have my doubts
ashen spindle
I still have my doubts for the general public after everyone seems to think nvidia are the saviors bc they made it so cheap..... while forgetting that turing just hiked the prices so much
3080 still 200$ more than 1080 lmao
shadow minnow
No 7nm 3090? 
ashen spindle
ofc not
pale sigil
i just dont trust amd
ashen spindle
you're not getting the a100 chip
shadow minnow
No 7nm no DP2.0 rtx 3090 disappointed me
ashen spindle
sure w/e
thick elk
Honestly looking at the announcment I'd say it's better than expected.
That being said take their performance numbers with a big grain of salt
vagrant marsh
we wait until amd releases and then the comparisments can begin
thick elk
Or at the very least for independent verification of their claims
fickle ore
@ashen spindle What annoys me are ppl who think Nvidia prices are a deal...... I see it in many discords atm.
People are just sheep
thick elk
If their claims are true it's a pretty good generational improvement for the same price
Unlike turing
fickle ore
same shit price like Turing
ashen spindle
this is what people dont see
fickle ore
^
and that performance is RT values in the best case scenario.
And nobody will use RT for atleast 3-5 years
thick elk
To me turning was a step where they added 0 performance just added an extra price teir on top
ampere is still a pretty decent upgrade going 1080 to 3070, be it over quite some time
fickle ore
Anyway its gonna be 800$- 900β¬ graphics card after taxes and AIB
fickle ore
Paying 1k for graphics card 
to play 1 game
thick elk
1070ti is not that different from the 1080
I don't see the problem, if people are willing to pay more, it means more expensive chips will be made.
There were some pretty expensive sli / crossfire setups back in the day. Those were even worse because crossfire / sli are just a pain to deal with.
Just for the record turing was just shit
this is the true step up from the 10 series
Not that I will buy it, I will probably just wait till I can get a decent upgrade from amd.
pale sigil
if i was still on 1080p i dont think i would have upgraded
shadow minnow
I am probably going from some 480 amd GPU in laptop to 3080 or 3090
fickle ore
@thick elk Ofc it's a stepup. As it should be. Every generation is. No need to get so hyped about generational leaps to pay 300$ extra
thick elk
true but the price hike was also one generation ago. As long as people realize the numbers got offset by 10 compared to the 1000 series it's fine.
pale sigil
thats reasonably priced
thick elk
My 1080 can get away with 1440p 100-110 fps in most of what I play but I do notice that the settings do have to be turned down quite a bit.
pale sigil
exactly
thick elk
For performance I am kind benefitting an upgrade, I also have to spend money I get from work on hardware and well I have to burn quite a bit, so money is not an issue.
I do generally want to move away from nvidia because of their linux driver situation being a source of annoyance.
crimson wigeon
Does he test Witchard 3 on it
ashen spindle
lets hope that means review NDAs are over more than an hour before release
benchmark with vsync on
wtf
shadow minnow
3080 and 3090 have exactly double the leaked core count
ashen spindle
yesn't
pale sigil
how bizare
sick creek
i'm gonna end up getting the 3090 because in typical apple nvidia fashion they gimped the ram on the regular models
not sure why every tech ceo is obsessed with being the modern day steve jobs
homie even did a "oh and we have 1 more thing"
vagrant marsh
true
ashen spindle
yes AiB models not FE cards
vagrant marsh
but why buy it right away?
remember the rtx 2000 series
the very first production line had some pretty faulty cards
sick creek
FE cards have way better support for water cooling unfortunately
ashen spindle
and the cores are double because they doubled the fp32 cores per SM, hence its exact double what leakers said, bc they only knew the SM count and assumed turing/a100 structure
and even then that got leaked, people were just cautious about that rumour
EK I think already said it will have blocks ready on launch
not sure if EK but one vendor has confirmed it
vagrant marsh
oh no
ea play is now available on steam
NVIDIA RTX IO: Enables rapid GPU-based loading and game asset decompression, accelerating input/output performance by up to 100x compared with hard drives and traditional storage APIs. In conjunction with Microsoftβs new DirectStorage for Windows API, RTX IO offloads dozens of CPU coresβ worth of work to the RTX GPU, improving frame rates and enabling near-instantaneous game loading.
special compression chip XD
ashen spindle
?
ofc you want to offload stuff like this on a console on a chip
why waste gpu performance on this on a console
vagrant marsh
ashen spindle
purpose built FPGA's are always better than general purpose...
vagrant marsh
still
microsoft's directstorage
for windows
nothing about linux
i dont have penguin
i have windu
dx12 2.0 cause 1.0 was so successful
half arse implimitation by game devs
like putting super behind it will make them do better
supah directX
thats the next one!
seriously
ultimate directx 12!!!!!
seriously, just let it die and move on to 13
shadow minnow
Rocket lake s when?
ashen spindle
why does the name matter so much to you
vagrant marsh
cause its a meme.
like putting ultimate behind it suddenly changes every dev that implements directx12 into not half arsing it
ashen spindle
at worst people ignore it, at best they fall for marketing, driving up how many people have never gpus so its makes more sense for devs to work on the api
vagrant marsh
maybe if vulkan was less succesful?
i seem to remember that vulkan caused a better improvement in fps than dx12
ashen spindle
then it means the vulkan implementation was better
vagrant marsh
probably
ashen spindle
you can just say x or y is better
hence we see rocket leake move up the dx stack and R6 go to vulkan and not dx12
vagrant marsh
i mean, maybe vulkan is unreal engine and directx12 is frostbite. maybe 1 is harder to work with than the other
ashen spindle
yes, but which one is easier to work with depends on what you do
thick elk
My understanding of dx vs vulkan on a technical side is the performance does not differ much, directx is faster to adopt new features, vulkan can be used more widely.
fickle ore
Both are good but u can't get lazy human beings study new shit
its just the way world is
why make games on d12 when dx11.3 games are good enough
Msoft keeps adding stuff on dx12 on purpose. One day there is enough stuff that might entice some developers.
ashen spindle
also you have devs with decades of dx experience and dx12 is so much more closer to hardware level, it's like they are as fresh as an intermediate dev working on dx11
so ofc it's not as fast adopted
wheat plume
runic geyser
wheat plume
the what?
crimson wigeon
the what?
wheat plume
the NAANNII?
wheat plume
ahhha
vagrant marsh
i mean, you enjoy people...who had the finances to buy a 1200+ euro graphics card, and are now...seeing that technology evolved and can now deliver that kind of power at a lesser price...
ahh, yea. 3 months
well, they will sell them at a loss for sure. the 3070 at only 500 quit being as powerful. the 3080 at only 300 more being twice as powerful
my next one will probably be a 3080 tho. but i am waiting to see a few months if the 20gb version comes out and go for that one, if it aint that much more expensive.
driver updates
how coinsidental
wait wut
world of warcraft got ray tracing?
how do you raytrace cartoon colors...
hmm.....
will i?
- not far off from the 700 they said it would be
that 1549 for the 3090. damn. thats twice the price of the 3080 tho
christ
vagrant marsh
by the time we reached the 6080, hte price will be 6080 as well XD
thats some bad rgb yo
it makes the thing look cheap
old marlin
XC3 looks good lol
vagrant marsh
xc3 looks generic
lets look at the marketplace if the 2080Ti is already being mass sold
today, today. today.
XD
thick elk
Mostly people trying to make a quick buck to not have it drop in value too much
vagrant marsh
alot of 2080Ti's put online for sale today
ohh
rtx 2080Ti xtreme for only 700 bucks
thats not bad
shadow minnow
rtx 2080Ti xtreme for only 700 bucks
@vagrant marsh compared to same performing 3070 for 500 it is bad
vagrant marsh
hahhaha
fuck it, im gonna do the same
im offering only 500 for all of em
lol
alot of them got minimal bidding limited to the asking price XD
priceless
interesting. alot of blower styles put up for sale
fuck off. offer too low
XD
hahahah
shadow minnow
I would buy 2080 ti for max 450 now
As 3070 will probably be quite better at ray tracing
ashen spindle
gddr*
thick elk
while gddr is more specific it would still be a subset of ddr memory.
vagrant marsh
i mean, if rtx IO can make it so that it enhances game loading speeds, then it seems to me it can be used as regular memory
volatile or not
ashen spindle
.. both are volatile
vagrant marsh
then there is no difference?
ashen spindle
G does stand for graphics, not persistent xD
we've been over this, one is optimized for latency one for throughput
gddr is right next to the chip, ddr needs to route through the board to the chip
vagrant marsh
no, but it now allows for the onboard GRAPHICS memory to atleast take over some of the functions of the ram
cause usually it goes over the ram and the cpu, while now its directly going through the graphics card
as simple as the graph can be
ashen spindle
yes that changes how the data is processed, it doesnt change what the vram does
like you couldn't stick gddr6x sticks into your dram slots bx the traces cant deal with PAM4
thick elk
Just to make a point, the ps4 uses gddr also as it's regular dram for the cpu
ashen spindle
so do both next gen consoles, im not saying they are 100% not related, obviously its both flash memory
but if you specify memory, why not either be accurate or say only memory
oof thats for 3090/80/70/2080ti
thick elk
qdr is dead sdr would be irrelevant so yea I guess so.
shadow minnow
As 3090 is out, Now I just have to hope rocket lake s getβs released at the end of this year and 11900k is good
ashen spindle
thick elk
DDR would however specify that it's not HBM
vagrant marsh
Well, i ordered some new memory and see if that is the problem with my pc. It still sometimes go in a startup loop. Strangely enough, i did found a work around. When it goes in a startup loop (it gets stuck when bios checks the ram) i simply reset the pc (front reset button) and that...resets the startup loop? And then it boots up normally.
Could also be that the motherboard is fried, that is why i need another set of memory to know for sure.
I hope its the memory sticks. I will simply return the faulty ones.
Replacing the motherboard is...problematic
thick elk
Usually if it's memory errors trying one of the two sticks is a good test
you can test both memory modules individually and both memory channels individually
vagrant marsh
Thats the problem. I did a full day of memtest on the memory and came back negative. Not a single error. It happens at bootup, something is interfering the bios from fetching the xmp profile on the memory.b
Without the xmp profile enabled, nothing is wrong.
thick elk
I also have a somewhat similar issue
if I have my pc turned off / disconnected from mains for a long time and my room is not very warm my memory will fail training or retry for a while and then succeed, rebooting works fine.
vagrant marsh
The problem is also inconsistent. Sometimes it happens 1 time after another when starting up the pc, sometimes it wont happen for 3 days.
thick elk
I never bothered to care much about it. I just pointed it at my memory is not stable on a cold system
vagrant marsh
Lol
thick elk
thing is my issue is consistent and there are no issues later on
Had it with a different set of memory as well.
ashen spindle
220W is awesome? a 1070 had 150W lmao
thick elk
As for memory stress testing, I don't know how comfortable you are with linux but using google's stressapptest is the gold standard for stability tests
It can find issues in 2 minutes were I won't find them in a long play session
ashen spindle
220W is barely below a 1080ti and that was on 18nm tsmc?
yeah so the lower perfoming samsung node throws away all the efficiency they gained from the uArch changes lul
thick elk
woo 5 whole watts compared to a not so great architecture and on a pretty big node shrink
vagrant marsh
I dont have linux so im not comfortable with it at all :p
shadow minnow
Jensen be like
thanks for crowdfunding AMPERE by buying one of our Turing cards
thick elk
I think amd does make strides on power, the rx5000 series is clearly more efficieny and the next gen consoles run on a lot lower power budget.
With more efficient I mean compared to their previous cards
ashen spindle
yeah compared to a gpu gen that got price hijacked like crazy
ashen spindle
nvidia basically increased prices by like 50% with turing, now they lowered them by 20-30%
so yeah, cheaper than turing sure
lower TDP than the others? sure, low TDP? lul ok
you're right
it will be better
thick elk
I expect amd to beat nvidia on value, however not in raw performance at the top.
ashen spindle
idk where you got that rumor from
but if amd could not make better cards, why would nvidia have rushed this generation to be out before amd
oh btw, anyone else noticed how they made 0 performance in normal rasterization, only on that arbirtray graph? the fps comparisons with 2000 was only with dlss/rtx
thick elk
there was this digital foundry comparisons which I think put the 3080 to about 60/70% above the 2080
ashen spindle
yeah but DF was only allowed to use selected nvidia titles to show relative performance
thick elk
not the worst selection of games from what I have seen, so not too cherry picked, they were however free to test how they wanted in terms of settings.
Certainly not perfect but it puts it in a ballpark
ashen spindle
sure, thats what nvidia says Β―_(γ)_/Β―
nvidia also said we have rtx everywhere now
vagrant marsh
And you dont believe em?
thick elk
Of course it would be foolish to take the digitalfoundry stuff as true but it is somewhat of an indication.
ashen spindle
I dont think he made inaccurate tests, but he still had to follow nvidias exact guidelines with settings and all
.. you're really dying on the hill that a nvidia marketing slide is that accurate, that it will be a 5% difference
god I wish I had that much confidence
oh god
and how is that measured
tf does 3x quieter mean
.. yes I have seen that much
thick elk
3 times quieter in marketing > reality: 9db less when having the microphone right next to the card at the spot that was loudest on the old card.
This only applies to the reference / fe card. Most people will not use in their systems.
ashen spindle
they also have lower TDP so they dont need a monster cooler
thick elk
Unless they decide to push the cards to the limit to beat nvidia again
ashen spindle
yeah then its still lower tdp most likely
and you got that info on navi performance from where?...
thick elk
there are some rumors going around about it matching the 2080ti rougly
official statement is just 50% more efficient than rdna1
the watt recommended is just based on the tdp and a very caucious estimate about the rest of the components and the quality of the spu
Also just because a card uses dual 8 pin does not mean it strictly needs it. Sadly the way the cards are designed it does require both to be plugged in
ashen spindle
and the fact that around 50% most PSUs are most efficent
thick elk
I don't think that factors in at all / I also don't think that's generally true / relevant as the psu also does not get near peak load most of the time.
mostly to account for somewhat shitty power supplies and power hungry cpu's
thick elk
fe pcb is pretty good nowadays from what I heard.
ashen spindle
it was for turing, we dont know if its the same quality for ampere yet
thick elk
We shall see
zenith sonnet
So, what did Leather Man say?
thick elk
realistically this type of questions can only be answered once reviews come out.
ashen spindle
yeah RMA in the us is actually a big factor, thank god for EU laws to protect customers
oh wait
the 3070 doesnt have gddr6x
just regular gddr6 lul
I didnt notice that
crimson wigeon
Great, is it also going to have a slice of slow RAM so we can have the 3.5 GB thing again
ashen spindle
haha doubtfull but would be funny
idk, if the leakers who had full specs of ampere in march say rdna2 will be good, I doubt they mean 3070 level only
zenith sonnet
It's only a matter of how big AMD can make RDNA2
But looking at the XSX, they can easily fit 36+ TFLOP into 300W
Sure
But the XSX is 130W CPU+GPU combined
And looking at Renoir is should be roughly 45W CPU and some 80W GPU
And those aren't even good bins
AMD should easily win perf/watt this round
We just don't know big they're going/can go
shadow minnow
Wondering how many games gonna run 4K 144 with 3090
ashen spindle
you do know we already have a Hotchips talk that details RDNA2...
like it wont get more technical than that until after release
well yeah it is, thats why its funny how many people cling to nvidia
I've been shitting on amd for years, hell I still do bc their marketing for rdna1 and all that was horrible
shadow minnow
I mean, it is usable with 8k
But for normal use with higher frame rates it is 4K card
ashen spindle
8k *with dlss
zenith sonnet
Also, just realised:
AMD is set to manufacture 10s of millions of console APU in the coming months, I fully expect them to break 100 million units some time in the second year of this console cycle
Even assuming a phenomenal 95% yield on console chips, AMD would be sitting on millions of amazingly good APUs that didn't make it, a little binning and cutting, and they've got quite the line-up
Now, the XSX APU is a 56 CU part, with 54 CUs enabled. If AMD were to keep the top bins for other purposes (think 56 CU functional with great voltage characteristics), they could have some few million bloody amazing chips
Van Gogh might just be disruptive, if they do it right
shadow minnow
It is imo definitely not overpowered for 4K 144
That is... unreasonable @prisma crag
thick elk
I'd say keeping the card for the next few years you would be looking at a lot more demanding games as well.
shadow minnow
Imo 3090 is for
1080 p 280-360
1440 200-240
4K 120-144
8k 60-75
Similar with 3080
But 3080 8k just max 60
ashen spindle
with DLSS in a supported title maybe
shadow minnow
Max useful 27β resolution for 95% of people 5k, for 80% 4K and for minority 1440p (to not see pixels on screen)
I want to go just 4K, eve spectrum 4K 144
zenith sonnet
First I wanna see a good UHD monitor
shadow minnow
monitors are overpriced and there aren't any hdmi 2.1 ones yet
@Capt.Nemo#3032 lg cx, eve spectrum
No but I plan to buy it
I would like to see QNED 27β 4K 1ms 240 as endgame one day
It is different and way better
Nope
Not small enough
Or QDLED I think is end game
Just If there was not burn in
QDLED should be perfect size, blacks, brightness, speed, price 
I would like to also get shiny new rocket lake s this Christmas so I can finish my build
I mean, under 1500β¬ gaming pc I would go ryzen, but over 1500β¬, intel for gaming
I9 10900k is best
And i5 10600k still better than anything amd can offer now
But we will see if that changes with zen 3 and rocket lake s
Rather aorus
Tbh
ashen spindle
yes the 90 is what they called titan
just marketed for gamers bc they pay the price anyway
ofc not but now it wont feature more performance really
oooor, they made rdna2 in cooperation with sony and microsoft and neither of them can really talk about it before it launches
negative, big navi is the first rdna2 launch they said
so its slated to be before console launch
given that fanboys ate the 20 series up at launch... yea that happens regardless of what amd does lmao
... where did we get the info from that FE uses any form of binned chips?
they used A/B chips or whatever, but AiBs also got both and they weren't binned a lot
nvidia using ampere before 2021 is bc amd is competetive lol
should all be RTX I wager
they cant afford to keep gtx around, with rdna2 having RT cores that are not sperate so they can do raytacing in the entire stack basically
yeah... great uArch sadly doesnt make the marketing better too, they could just fuck it up again like last year
and worse value than pascal
jensen literally needed to personally tell people to finally upgrade from pascal
you mean Thermi? 
we've seen that it doesn't already...
yes
we've seen it in an amd video months ago
even the 7 had only 295W lul
ashen spindle
well thats the issue with FOSS
pale sigil
sick creek
aight?
frozen anchor
So no 3060 only 3070 I guess
sick creek
they know the masses are poor, why would they kill products before launch by making cheaper cards available π
there will be a 3060, just not at launch
it will probably just be a 3070 that didnt make the cut
frozen anchor
I'm thinking about getting a 3070 but I might as well just get a whole new PC aswell since mine is pretty old at this point
So I might just wait till new amd CPUs are out
sick creek
considering its pci-e gen4, you'll probably need to anyway
RIP to all the people who ragged on x570
real question now is... who will they get upset at more? nvidia or amd
sacred shoal
@ashen spindle isn't it still considered FOSS if it's released with a copyleft license
misty pier
true quest
skip the 3060 altogether
shadow minnow
Why settle for it when you can get quadro
vagrant marsh
I wonder if we get cheaper cards as well
ashen spindle
considering its pci-e gen4, you'll probably need to anyway
a 2080 ti still gets like 96% performance on a x8 pcie3 slot, so you dont need pcie4 for the 30xx
maybe theyre are some titles where its mostly bandwidth bound? but thats unlikely
old marlin
pcie gen 3 could be a bandwidth limitation if you need to shovel a ton of stuff on the VRAM fast
since the GPU can now directly get stuff from NVME SSDs
shadow minnow
Not now, that feature should be released next year π
ashen spindle
Imo that makes it even more likely you donβt need pice4, direct access = even 3 has faster access than 4 through ram
shadow minnow
Also, 5700xt is imo so bad deal now, as 3070 is so much better, for not that much extra money.
reef patrol
The 3070 isn't available. So that statement is... kind of wrong π
old marlin
no people know how the market will exactly look like before there are any price public for aftermarket models
reef patrol
If the 5700XT is still priced like that when 3070 stock becomes available, then we can talk π
thick elk
when th 3070 will be available th 5700xt will drop in price
ashen spindle
This just in, old gpu became worse after new generation came out, everyone is shocked, more at 11
old marlin
I am surprised that nvidia seems to put more and more care tin their reference cards though
they put no care in aftermarket coolers on the reference PCBs though
odd shape
odd power connector position and angle
ashen spindle
I'm not, they dont want reviews to start by saying the card thermal throttled
just look at 5700xt reviews, how hot that card got
on release
now add 100W for 3090
ashen spindle
they were also hotter in reviews that used k-types to show heat so thats not the only reason :p
https://twitter.com/HardwareUnboxed/status/1301048043232358400 see, nvidia knows how to control their perceived image quite well
pale sigil
I thought this was pretty obvious
I do like digital foundry a lot, just not the conditions they were made to follow
ashen spindle
yea its obvious to the minority of people
DF sometimes goes deeper into tech, but based on their general content I dont see the majority of their audience understanding whats the difference to a review
pale sigil
That twitter thread is awful
thick elk
This goes to show that if you are a media outlet like digital foundry you really should not do this type of content.
Makes it very hard for some people to see the difference.
Still I do think it's a little bit above nvidia numbers in terms of reliability, nvidia controlled most but not all variables, I'd put it at a decent best-case were I expect the total average to be about 10-15% lower. They did pick larger games and settings that are not all that weird + they could still decide were in game they wanted to test.
pale sigil
ashen spindle
it's the "please dont rma we even added more coolers" version
vagrant marsh
Did you know that UPS got a trademark on the color brown?
pale sigil
Nope
ashen spindle
telekom got one for their magenta
vagrant marsh
They say in their mail XD my package got delivered. The ups name, logo and the color brown are trademarked
ashen spindle
or t mobile or w/e its called there
vagrant marsh
XD
pale sigil
Hmm maybe I should use "UPS brown"
vagrant marsh
Haha
thick elk
Just remember trademarks are pretty limited, so if a restaurant used the same color for example it would not be a problem.
ashen spindle
https://videocardz.com/newz/nvidia-geforce-rtx-3070-ti-spotted-with-16gb-gddr6-memory Wonder of Lenovo just made an error or if TI versions get double ram this time, could make Sense
pale sigil
oh fuck me
ashen spindle
?
pale sigil
Just, ugh why cant we just have 3 gpus
ashen spindle
.. yes lets have less options bc if you want more of one part just pay 300$ more for the next tier
???
thick elk
Kind of weird, to give it 16gb vram, kind of messes the whole product stack up in terms of vram numbers
which is something the average consumer still watches
ashen spindle
now thats just stupid reasoning
ashen spindle
so they should just never have made the super variants of turing either?
pale sigil
they should have been a thing from the start, instead of rebranding etc
ashen spindle
yeah so everyone will complain about no stock, great idea
thick elk
turing as an architecture was kind of a dud / stopgap that overstayed it's welcome quite a bit.
pale sigil
Why cant they just offer the best for the price bracket instead of having middle men
Pisses me off that in a few months your gpu becomes a lower tier variant and costs the same usually
ashen spindle
then wait with your purchase Β―_(γ)_/Β―
thick elk
Price drops in gpu land are quite common especially if the competitior does something
thick elk
Would not surprise me if rdna2 prompts nvidia to adjust pricing a bit as well.
They did calculate it in but they don't know the exact numbers / pricing.
plush summit
Meanwhile I've lost all hope in AMD
halcyon moss
he's sleep deprived, thats why :p
does anyone else manage to scrape of the soft touch rubber like layer on keyboard/mice when trying to clean it? I always think its just a big amount of accumulated dirt and grime, so I rub harder and then this layer comes of
fickle ore
When are those "true" pcie 4.0 controllers coming e.g phison 18/19? And released in proper SSD's
ashen spindle
the new samsung ssd should have one, 980 pro iirc? it got shown recently and should be out soon-ish
ashen spindle
no samsung uses their own
fickle ore
oh
fickle ore
yeah it is not true 4.0 gen I think
cause true gen 4.0 uses nvme 1.4 spec
tho it has 1 mil IOPS so hmm
thick elk
thos iops are probably at a pretty high queue depth so not too important in a lot of cases
ashen spindle
its adding the samsung tax anyway so Β―_(γ)_/Β― idk if I'd go for it anyway
fickle ore
Building a new system along with zen 3 so got still time
thick elk
It's their pro series which in general has always been rather expensive.
fickle ore
putting all in Lian LI lancool II mesh
@ashen spindle kinda decided already to go with AMD due to the fact MSI is on by default. Can't be arsed turning it on every time I upgrade the driver
ashen spindle
lul I mean unless you get a deal I dont see any reason to go intel anyway
fickle ore
Nono I meant for the graphics card
ashen spindle
oh
fickle ore
for Nvidia you need to use the program to turn on MSI every time after driver update and reboot your PC
Hate that
ashen spindle
I mean I want to say good choice, but in the end thats only true if the drivers work much better
in the end, if you need to constantly fix something, thats the same time spend as using your tool
but I am praying the drivers wont have as many issues
ashen spindle
depends, do you play any games that support it?
ashen spindle
ye, there are rumors a future version will work on any game that has TAA but thats not confirmed yet
fickle ore
Nvidia features are always the same crap, need to be implemented by the gamedevs. Only the big studio SP games usually implement them anyway
UE 5.0 deprecated PHYSX π
ashen spindle
ehh, depends, I mean for hairworks it makes sense that it can only work if implemented
fickle ore
What lower latency, you mean the nvidia reflex?
fickle ore
that's just basically improved "low latency ULTRA" and only works in GPU bound scenario, in which no competitive players ever are
So its just for SP games at high settings
I guess could be nice for a mmorpg
ashen spindle
I mean, if im not playing a csgo like game, I still like better input at high quality graphics
fickle ore
Yeah for sure but idk how much improvement can you get out it vs low latency ultra? 10% chance for - 1 to 2 ms?
ashen spindle
idk nobody has the cards to test yet so we only have nvidia numbers and I dont look that them lol
fickle ore
IIRC AMD's implementation of the low latency mode is better as per their inventor (he talked about it on guru)
pale sigil
Won't most reviewers have them but are under nda
ashen spindle
ehh, bit biased as he's from AMD
fickle ore
True but he designed it and nvidia followed suit
ashen spindle
tbf they were fairly quick to follow up so it makes sense if its a bit worse
wheat plume
it seems CUDA is the reason this upgrade is 2GB large.....
400MB and its not even close to done downloading
ashen spindle
.. one of them is not what I expect as speaker for TGL launch
ashen spindle
wheat plume
SHES GOT THE KAREN HAIRCUT TOO
vagrant marsh
Short but spicyyy
chrome estuary
Also of course there's a "Corporate Vice Manager" and a "Senior Vice Manager"
vagrant marsh
Cause having 1 vice manager aint enough money thrown away
Now thats a tight fit
Might have to move that front fan to the back
Alright. It auto detected the xmp profile which is a good start
16 gigs running at 3600. good
ashen spindle
20 minutes till TGL, lets see if its mobile only or if they have some surprises
vagrant marsh
Whats tgl?
Wow. With this new memory my pc starts up quietly. With the other memory the fans went to 100% before settling down.
vagrant marsh
I know
pale sigil
Do you not fully shutdown windows?
vagrant marsh
So maybe it is the memory that isnt working right
No, i do turn off the pc when i turn it off. No sleep mode or something
reef tundra
The karen will speak as the manager to the manager with audience. After that not even the CEO will unhear the complaint about how rude the maneger was hearing her complaint on how rude the worker was while she was complaining about the pizza being too hot and burning her mouth or some other big think ideas.
Also most executive titles are bs i wanna feel important so i am a hygyne sanitation specialist at the local mac'ers
vagrant marsh
Updated my bios to the latest version as well
Appearently my bios can now access files on the drives
frozen anchor
whats that supposed to be used for?
vagrant marsh
Maybe update bios from the drive directly? To load up a downloaded bios profile?
Who knows
vagrant marsh
i will continue to call it bios whether you like it or not, young man.
But this settles it. My memory is busted. I will run it for a few more days and then return it under warrenty
vagrant marsh
squirting
ashen spindle
great, intel cant even make a smooth video playing for the TGL launch
this feels like 24fps
reef tundra
Intel? what about outtel?
Also usually you want the memory to use the closest slots to the CPU (interlaced for dual channel). But some mobos are very specific about this
vagrant marsh
i think the general concensus is that you use 2 and 4, not 1 and 3.
unless that doesnt matter on modern motherboards
reef tundra
Less tracelength means less inpedence so higher clockspeeds are achieveble, π
but dont know how much that would benefit
also what brand is that
looks like some offbrand chinese fake sticks
ashen spindle
patriot viper? looks like some snake design
reef tundra
overreporting itself causing it to shit the bed when actuallly having to allocate the nonexistant area
ashen spindle
... yes
vagrant marsh
ashen spindle
thats why I guessed its patrio viper ram kit xD
vagrant marsh
you know. the logo of a snake with its mouth open
yea, its that one
only 80 bucks 3600 cl17. but if this memory proves more stable than my last memory i will try to OC it
reef tundra
Lifetime warrenty?
reef tundra
Usually if they work they will keep working untill the end of time
vagrant marsh
ohh
reef tundra
so they can affort to give lifetime warrenty without many risks
ashen spindle
hey nice, Xe has a copy of rtx voice
but god that was the most cringe way to showcase that lmao
reef tundra
Cringe
crimson wigeon
Love me some crinch
vagrant marsh
man, this new memory is only 80 euro but performs about the same. my bl3 benchmark didnt drop. the other memory was 144 euro when i bought it. man, i way overpaid for it.
guess rgb is expensive
XD
reef tundra
vagrant marsh
intel is having its reveal now?
reef tundra
intel is being itself and dropped a core so they can raise the price
That seems to be a little cheap, as in from the rest pile after the binning process cheap
vagrant marsh
wait what, dropped a core what?
ashen spindle
you know it's shit if they have diplo take up a part to convey that a laptop is usefull lmao
reef tundra
taps furiously while nothing changes
ashen spindle
vagrant marsh
haha
reef tundra
Technically better than the abbacus in 4d linear matrix transfoms
pale sigil
livestream?
ashen spindle
"look at how smooth the steam is compared to the competition" yeah girl, I can really see the smoothness of that stream ON YOUR LAGGY STREAM
pale sigil
link>?
pale sigil
cba
ashen spindle
go to anandtech for the liveblog then
vagrant marsh
where?