|
Riflen posted:Firstly, I wouldn't be surprised if Nvidia have negotiated at least a period of exclusivity on GDDR6X, given they collaborated to develop it. No other company ever used GDDR5X either for example. Has AMD confirmed move away from HBM2 or is this just sort of a logical conclusion based on HBM2 cost/performance vs GDDR6?
|
#
?
Sep 2, 2020 16:19
|
|
|
|
| # ? Jun 4, 2024 00:23 |
|
|
8-bit Miniboss posted:Don't use email notifications. It's always slow. Text is much more immediate. They don't send text messages to my carrier 
|
|
#
?
Sep 2, 2020 16:24
|
|
|
Nfcknblvbl posted:I have been using nowinstock.net before but apparently there's a 10+min lag with them so that's why I'm trying out the extension I mentioned. I can't find the name of this extension in your earlier posts, and it is possible that it's not just due to my brain being bad. Could you remind?
|
|
#
?
Sep 2, 2020 16:25
|
|
|
Subjunctive posted:I can't find the name of this extension in your earlier posts, and it is possible that it's not just due to my brain being bad. Could you remind? Page Monitor. You can make it so it monitors a single element in the page to avoid false alarms.
|
|
#
?
Sep 2, 2020 16:26
|
|
|
HalloKitty posted:OK holy poo poo that was a lot of posts I didn't even bother to read, this thread exploded, and it's obvious why. With the exception of the 3090, it went WAY beyond expectations.
|
|
#
?
Sep 2, 2020 16:27
|
|
|
Nfcknblvbl posted:They don't send text messages to my carrier They don't for me either which is why they suggest using TextNow. Works just as well.
|
|
#
?
Sep 2, 2020 16:28
|
|
|
Generic Monk posted:i think the supers mainly came out because the 20xx series was essentially a dud on launch in terms of raster performance and it was affecting their sales. looks like they've hit it out of the park with this one so there's probably not much need to do that. excellent marketing job cranking up the prices with the last gen making people actually kiss the ground at their feet when they didn't this gen, too. Maybe, but the TI line was basically the same thing as super refreshes before ti's got rolled into release. I don't think yearly arch changes are realistic anymore and going 24 months without a new line would be a really long time and a lot of awkward quarterly meetings with flat/declining sales if people are able to just sit pat with their 3070/3080s. The thought that there may be super refreshes with higher memory limits/mediocre performance gains seems realistic, as is the idea of the Turing refresh where they basically bump off sagging models with a refresh (that's hopefully a bit more power efficient). I think a super refresh is also a good way of goosing sales when you have a more profitable production line, and I don't think NVidia would want to pass on that. I'm not saying it's a reason to wait for the refresh or anything, any more than the immutable law of graphics cards will always be "There's something better coming out sometime in the next 12 months".
|
|
#
?
Sep 2, 2020 16:28
|
|
|
Zedsdeadbaby posted:With the exception of the 3090, it went WAY beyond expectations. Usual disclaimer of "wait for real reviews" applies but yeah, unless nvidia are lying their balls off the 3070 and 3080 look great and the price hike rumors were false. The 3090 is PROBABLY like 99% there to rip off XTREME GAMERZ who want maximum e-peen, and 1% niche users who have a legitimate use case for that sort of thing.
|
|
#
?
Sep 2, 2020 16:31
|
|
|
Smoke on the street is that the Nvidia mandated sell time is 2:00PM BST
|
|
#
?
Sep 2, 2020 16:32
|
|
|
KYOON GRIFFEY JR posted:Has AMD confirmed move away from HBM2 or is this just sort of a logical conclusion based on HBM2 cost/performance vs GDDR6? Apart from the mobile 5600M, no RDNA GPU has been shown that uses HBM. The GPU design is tied rather closely to the memory technology. It's not impossible for the 6000 series to use HBM, I'm only guessing. I was just pointing out that they could technically compete with 3080 on memory bandwidth using GDDR6. Having said that, comparing a single stat on two different architectures often doesn't yield much useful information. It may be that one can require more memory bandwidth to reach a similar level of performance.
|
|
#
?
Sep 2, 2020 16:46
|
|
|
KYOON GRIFFEY JR posted:Has AMD confirmed move away from HBM2 or is this just sort of a logical conclusion based on HBM2 cost/performance vs GDDR6? It's mostly a conclusion based on them not using it for the 5-series, nor in the XBox/PS5. They might have a halo card that uses it just so they can claim they have higher memory bandwidth than the 3090, but the rank-and-file cards that are actually intended to be bought by real people will be GDDR6, almost assuredly.
|
|
#
?
Sep 2, 2020 16:46
|
|
|
hatty posted:Smoke on the street is that the Nvidia mandated sell time is 2:00PM BST
|
|
#
?
Sep 2, 2020 16:56
|
|
|
Wiggly Wayne DDS posted:yeah i'm seeing that on uk retailer sites So 9AM EST?
|
|
#
?
Sep 2, 2020 16:57
|
|
|
NoDamage posted:Um, yeah? Putting together a deep learning rig using a couple of consumer-grade GPUs is not exactly uncommon these days. doesn't work in data centers. You can do this for workstations, which is why Titans exist as a baby quadros/teslas, but it hits a limit pretty quickly.
|
|
#
?
Sep 2, 2020 16:57
|
|
|
KYOON GRIFFEY JR posted:Has AMD confirmed move away from HBM2 or is this just sort of a logical conclusion based on HBM2 cost/performance vs GDDR6? HBM is not cost competitive due to the huge honking interposer.
|
|
#
?
Sep 2, 2020 16:58
|
|
|
jisforjosh posted:So 9AM EST? So 9am on the 17th?
|
|
#
?
Sep 2, 2020 17:04
|
|
|
So with regards to gaming, does more VRAM mean anything (at least for 1440p and below)? Given the minor bump of 2GB on 3080, I'm assuming now, and from what I understand, RTX IO (Nvidia's DirectStorage) basically just adds as a compression/decompression layer to improve storage streaming which should work independent of VRAM (size), right? Or will RTX IO/DS also buffer more storage as a result or something? The 3070ti VRAM rumour just made me wonder.
|
|
#
?
Sep 2, 2020 17:15
|
|
|
Malcolm XML posted:HBM is not cost competitive due to the huge honking interposer. yeah, i was seeing cost estimates of twenty bucks for the interposer which is pretty significant
|
|
#
?
Sep 2, 2020 17:17
|
|
|
ufarn posted:So with regards to gaming, does more VRAM mean anything (at least for 1440p and below)? Given the minor bump of 2GB on 3080, I'm assuming now, and from what I understand, RTX IO (Nvidia's DirectStorage) basically just adds as a compression/decompression layer to improve storage streaming which should work independent of VRAM (size), right? For gaming, VRAM is not like RAM in a performance sense. With RAM, it's usually a function of what other stuff you have running in your PC, and the impact of paging being a huge hit. With VRAM its much more controlled and predictable, and to avoid paging you have tons of options like sliding view distance or texture size down slightly which can make a huge difference in VRAM usage for potentially only small impacts on visual quality. Having more VRAM than you need is usually no benefit whatsoever, and right now very few games can even get above 6GB at max settings even at 4K, and usually its trivial to cut that usage down. The few that do suck up VRAM are usually badly optimized, which isn't really a thing that is easy to try to solve with hardware in general. Now will we see the VRAM usage increase? Probably, but the impact is significantly reduced the further below 4K you get. Don't try to future proof with a purchase today. If VRAM is a driving factor you'd probably do a lot better buying a cheaper card now and using that money when you start seeing games that really have an impact with lower memory.
|
|
#
?
Sep 2, 2020 17:28
|
|
|
jisforjosh posted:So 9AM EST? Sickening posted:So 9am on the 17th?
|
|
#
?
Sep 2, 2020 17:30
|
|
|
8-bit Miniboss posted:They don't for me either which is why they suggest using TextNow. Works just as well. Yay, thanks!
|
|
#
?
Sep 2, 2020 17:36
|
|
|
Wiggly Wayne DDS posted:online orders are only being accepted from 10 minutes after opening, so don't touch any of the sites then Can you explain this one a bit more? Seems like I am missing some critical info.
|
|
#
?
Sep 2, 2020 17:37
|
|
|
hatty posted:Smoke on the street is that the Nvidia mandated sell time is 2:00PM BST
|
|
#
?
Sep 2, 2020 17:40
|
|
|
ufarn posted:So with regards to gaming, does more VRAM mean anything (at least for 1440p and below)? Given the minor bump of 2GB on 3080, I'm assuming now, and from what I understand, RTX IO (Nvidia's DirectStorage) basically just adds as a compression/decompression layer to improve storage streaming which should work independent of VRAM (size), right? At 1440p or below 8GB VRAM is almost certainly going to be overkill for quite some time. I wouldn't worry about it. Also the 3070Ti thing from Lenovo is more likely stray voltage from them prepping a variety of entries prior to NVidia making final decisions about names, models, etc. The same "confirmation" also had the 3080 and 3090 with GDDR6, rather than GDDR6X VRAM, so I would put precisely zero stock in it until and unless confirmed by other vendors--which you'd expect they'd have already done at this point. I mean, I think a 3070Ti/Super will probably exist at some point. Just not right now.
|
|
#
?
Sep 2, 2020 17:42
|
|
|
Combat Pretzel posted:British standard time? What sort of moon timezone is that? British Summer Time, so UTC/GMT +1 Pesky daylight savings time... vv To be fair, a fair few countries use DST, including almost all of the EU. CET becomes CEST HalloKitty fucked around with this message at 18:35 on Sep 2, 2020 |
|
#
?
Sep 2, 2020 17:43
|
|
|
Combat Pretzel posted:British standard time? What sort of moon timezone is that? The Brits couldn't stand letting the US have a special-snowflake shifting summer/winter time mess, so they decided to close the Time Zone Gap and invent British Summer Time. Which makes even less sense for them than it does for the US, but what can you do?
|
|
#
?
Sep 2, 2020 17:45
|
|
|
With the 1080 ti, people mathed out that it actually turned out to give more performance per dollar/euro than anything else. What are the odds this is happening again? If the performance and prices are as estimated this is looking exceptional.
|
|
#
?
Sep 2, 2020 17:45
|
|
|
We don't really know because they are going out of their way to obfuscate the performance of their own cards in like 4 different ways
|
|
#
?
Sep 2, 2020 17:47
|
|
|
BurritoJustice posted:What an unnecessary low shot at DF. Not the first time HU have slung poo poo at better reviewers without solid basis. He could've framed this in a way that is less accusatory and made a more compelling point Yeah agreed. While I'm not a fan of DF's 'sponsored' videos (and videos that aren't that veer into advertorial at times imo -eg: Star Citizen), they're at least upfront about it and in this video specifically they reference that directly "...and we've done sponsored Nvidia videos recently!". Leadbetter was perfectly candid in how constrained they were by Nvidia and this wasn't a review. In a similar vein, I made the mistake of clicking on a Moore's Law video on the release and he basically outright accuses them of being a Nvidia shill, ffs Generic Monk posted:i think they just like to be nice to everyone and let The Almighty Framerate Line do the heavy criticism, which is fine by me I think the lawsuit threatened by the Redout devs for DF being slightly off with their pixel counting had a hand in tempering the tone of their critique from what I've seen of them over the years too. Happy_Misanthrope fucked around with this message at 17:57 on Sep 2, 2020 |
|
#
?
Sep 2, 2020 17:51
|
|
|
Combat Pretzel posted:British standard time? What sort of moon timezone is that? 2PM BST Noon at the Cape Verde Islands, midnight in the Salomon Islands. These timezones are among the most sparsely populated timezones in the world. Even the half-hour time zones have more people living in them.
|
|
#
?
Sep 2, 2020 17:51
|
|
|
Vir posted:half-hour time zones What lunacy is this?
|
|
#
?
Sep 2, 2020 17:53
|
|
|
Happy_Misanthrope posted:In a similar vein, I made the mistake of clicking on a Moore's Law video on the release and he basically outright accuses them of being a Nvidia shill, ffs Nvidia isn't exactly doing its best to be transparent, to be fair. In fact this is probably the least transparent GPU launch I can remember but I would be interested to hear if people can think of others.
|
|
#
?
Sep 2, 2020 17:53
|
|
|
Sickening posted:Can you explain this one a bit more? Seems like I am missing some critical info.
|
|
#
?
Sep 2, 2020 17:54
|
|
|
Mindblast posted:With the 1080 ti, people mathed out that it actually turned out to give more performance per dollar/euro than anything else. What are the odds this is happening again? If the performance and prices are as estimated this is looking exceptional. For the 3090? Absolutely not. It's over twice the price as the 3080, and nowhere near twice the performance for gaming loads. If you're a ML geek, things might be different, especially since with 24GB VRAM you can use data sets that you simply couldn't fit into the 3080's 10GB. But for gaming, no, it's on the wrong end of the price:performance curve, as halo products normally are. The 3070 and 3080 look like they're likely to be pretty close in price:performance, so you can pick whichever one fits your budget/needs better without worrying overly much about it.
|
|
#
?
Sep 2, 2020 17:55
|
|
|
Taima posted:Nvidia isn't exactly doing its best to be transparent, to be fair. In fact this is probably the least transparent GPU launch I can remember but I would be interested to hear if people can think of others. Sure, but this guy is one of the last people to feel justified in critiquing the competence of any other channel. DF's videos are actually watchable and are usually more than "here's what I've read on forums recently" with clueless speculation added in.
|
|
#
?
Sep 2, 2020 17:56
|
|
|
Enos Cabell posted:What lunacy is this?
|
|
#
?
Sep 2, 2020 17:58
|
|
|
Wiggly Wayne DDS posted:the order system becomes active on the backend at 14:00, but there'll be a queue put in place for anyone trying to access it for the first 10 minutes to delay bots. if you access it at 14:10 or onwards you'll bypass it, so don't touch the sites for the first 10 minutes If you know this to be true, doubtfully the bots also know this. And since they're also almost certainly running off of AWS/Azure instances, they can afford to simply open two connections from different IPs, one at 14:00:01 and one at 14:10:01 just to ensure they get in there first, anyhow. A delay like that is utterly pointless, and more likely to frustrate legitimate buyers than bots. Are there places that are actually stating they're putting in a 10 minute delay, or is this just rumors?
|
|
#
?
Sep 2, 2020 17:59
|
|
|
DrDork posted:If you know this to be true, doubtfully the bots also know this. And since they're also almost certainly running off of AWS/Azure instances, they can afford to simply open two connections from different IPs, one at 14:00:01 and one at 14:10:01 just to ensure they get in there first, anyhow. A delay like that is utterly pointless, and more likely to frustrate legitimate buyers than bots. To be honest this feels like a rumor you spread to make it more likely you get an order in.
|
|
#
?
Sep 2, 2020 18:00
|
|
|
That's good to hear about the 16GB ti. I think 8GB may be cutting it close a bit for some titles going forward. Also hope they don't do something idiotic with the 3060 and give it 6gb. 8GB, min for any card in 2020.
|
|
#
?
Sep 2, 2020 18:00
|
|
|
|
| # ? Jun 4, 2024 00:23 |
|
|
Happy_Misanthrope posted:That's good to hear about the 16GB ti. I think 8GB may be cutting it close a bit for some titles going forward. Also hope they don't do something idiotic with the 3060 and give it 6gb. 8GB, min for any card in 2020. The 2060Super had 8GB, so based on what we've seen so far it would be surprising to see them release a potential 3060 slot card without the same
|
|
#
?
Sep 2, 2020 18:02
|
|
