1180's here in two months!

[x.momo.x]

Nvidia Turing: GTX 1180 specs leaked, plus everything we know so far

 

[reactor_sa]

Your maths is wrong.

Autumn is September (northern hemisphere), so 9-5= 4. 4 months to September.

 

[z3llin]

This article is reporting what was said in another article which does not have validated sources. So its just more churning of the rumour mill.
The next foreseeable event where Nvidia could make a big media splash would be Computex in June. If they are releasing in the next quarter there should be news about it there.

 

[JollyJamma]

This article is reporting what was said in another article which does not have validated sources. So its just more churning of the rumour mill.
The next foreseeable event where Nvidia could make a big media splash would be Computex in June. If they are releasing in the next quarter there should be news about it there.

I agree with this since Gamers Nexus has largely put out the same statement. Nvidia may launch consumer products in June but it is seen as more likely Q4 of 2018.

 

[Zinc]

I agree with this since Gamers Nexus has largely put out the same statement. Nvidia may launch consumer products in June but it is seen as more likely Q4 of 2018.

And looking at the leaked specs, it doesn't seem all that wonderful. Although, most spec comparisons were between the 1080 Ti and the 1180/2080 so not really apples for apples.

 

[Off-The-Chart]

wait what? 1180s?

I am still on 980... when did we get 1080s?

 

[JollyJamma]

And looking at the leaked specs, it doesn't seem all that wonderful. Although, most spec comparisons were between the 1080 Ti and the 1180/2080 so not really apples for apples.

I wouldn't pay much attention to the rumours. Wait for a while. It looks like it's going to be GDDR6 but they are still refining it.

I'd be very hesitant about early adoption of GDDR6. Also, the launch date seems to be end of May.

 

[x.momo.x]

Nvidia Turing: GTX 1180 possible prototype spotted, plus everything we know so far

 

[Ron Burgundy]

what's the point of following all these speculation and rumours, best to just ignore till something official comes along

 

[Mo Facta]

My opinion is we are in a slump at the moment with computational technology progress in general. NVIDIA's CEO has already stated that next gen GPUs are not coming for a long while. Meanwhile, Intel is struggling with sub-10nm lithography and their Computex demo was proven to be a farce with some repurposed 28-core Xeon "re-branded" as a new consumer CPU at 5GHz. My guess is we're seeing Moore's Law and other scientific limitations come into effect. Plus there's the cost of manufacture vs market battle.

Furthermore, we are kind of in a dubious zone when it comes to consumer upgrade paths. Sure, PCIe is not looking to change any time soon but what about CPU sockets? Intel's roadmap seems pretty set on the LGA1151 socket well into 2019. Who knows when they'll switch on us and then it's a whole new ball game of upgrading. Skylake-X is also on a perpetual course until further notice on the LGA2066 platform. Who know's what Kaby Lake-X is going to be. It's an interesting time for computing. One thing's for sure, though, at some point if you want to upgrade to the latest and greatest from a previous single generation, a system-wide upgrade will be in order. Then again, what's new? But I don't think things have been so obscure for a long while.

Also, Volta is nebulous and seems scrapped, or maybe not. Who knows? So now we have no idea what NVIDIA are going to do.

I say just stay patient and wait until stability materialises and make the best upgrade choice you can. In the meantime your 1080Ti will do just fine

That's my 2c.

Cheers

 

[KZN_86]

NVIDIA GeForce 11 Series and GeForce GTX 1180 Revealed By Lenovo

 

[SCHUMI4EVER]

My opinion is we are in a slump at the moment with computational technology progress in general. NVIDIA's CEO has already stated that next gen GPUs are not coming for a long while. Meanwhile, Intel is struggling with sub-10nm lithography and their Computex demo was proven to be a farce with some repurposed 28-core Xeon "re-branded" as a new consumer CPU at 5GHz. My guess is we're seeing Moore's Law and other scientific limitations come into effect. Plus there's the cost of manufacture vs market battle.

Furthermore, we are kind of in a dubious zone when it comes to consumer upgrade paths. Sure, PCIe is not looking to change any time soon but what about CPU sockets? Intel's roadmap seems pretty set on the LGA1151 socket well into 2019. Who knows when they'll switch on us and then it's a whole new ball game of upgrading. Skylake-X is also on a perpetual course until further notice on the LGA2066 platform. Who know's what Kaby Lake-X is going to be. It's an interesting time for computing. One thing's for sure, though, at some point if you want to upgrade to the latest and greatest from a previous single generation, a system-wide upgrade will be in order. Then again, what's new? But I don't think things have been so obscure for a long while.

Also, Volta is nebulous and seems scrapped, or maybe not. Who knows? So now we have no idea what NVIDIA are going to do.

I say just stay patient and wait until stability materialises and make the best upgrade choice you can. In the meantime your 1080Ti will do just fine

That's my 2c.

Cheers

Make up your mind. Are we in a slump or is it an interesting time?
You can't start with the one, then gravitate toward the other and then talk yourself all way around the bush as well.

 

[Mo Facta]

Make up your mind. Are we in a slump or is it an interesting time?
You can't start with the one, then gravitate toward the other and then talk yourself all way around the bush as well.

A slump can't be interesting? I guess it depends on your point of view, doesn't it?

If you disagree with me, then just address my points. No need to be a raspy grump about it.

Cheers

 

[SCHUMI4EVER]

A slump can't be interesting? I guess it depends on your point of view, doesn't it?

If you disagree with me, then just address my points. No need to be a raspy grump about it.

Cheers

There's nothing to address because aside from the contradictions and the suggestion people hang onto their 1080Ti's you don't say anything, just make general statements or non-statements. It's like you are practicing to be a CEO.

 

[Mo Facta]

There's nothing to address because aside from the contradictions and the suggestion people hang onto their 1080Ti's you don't say anything, just make general statements or non-statements. It's like you are practicing to be a CEO.

Dude, what's your problem? I opened my post with "In my opinion". Ad hominem, much? I'm trying to be civil but you're just being a prick.

If you want to debate or disagree with what I said, then great. Otherwise, what is the point of your posts besides having a go at me and then attacking my character? If you don't agree with what I said then come back with a rebuttle. I'm not afraid of being wrong. Otherwise, keep it to yourself without all the condescension.

 

[AAAH NUUU BRU]

wait what? 1180s?

I am still on 980... when did we get 1080s?

Come to my house, i have a room full of them if you want to see what they look like

 

[D3TTOL]

There's nothing to address because aside from the contradictions and the suggestion people hang onto their 1080Ti's you don't say anything, just make general statements or non-statements. It's like you are practicing to be a CEO.

Dude, what's your problem? I opened my post with "In my opinion". Ad hominem, much? I'm trying to be civil but you're just being a prick.

If you want to debate or disagree with what I said, then great. Otherwise, what is the point of your posts besides having a go at me and then attacking my character? If you don't agree with what I said then come back with a rebuttle. I'm not afraid of being wrong. Otherwise, keep it to yourself without all the condescension.

 

[Flex]

LoL

 

[Mo Facta]

 

[Gouhan]

My opinion is we are in a slump at the moment with computational technology progress in general. NVIDIA's CEO has already stated that next gen GPUs are not coming for a long while. Meanwhile, Intel is struggling with sub-10nm lithography and their Computex demo was proven to be a farce with some repurposed 28-core Xeon "re-branded" as a new consumer CPU at 5GHz. My guess is we're seeing Moore's Law and other scientific limitations come into effect. Plus there's the cost of manufacture vs market battle.

Furthermore, we are kind of in a dubious zone when it comes to consumer upgrade paths. Sure, PCIe is not looking to change any time soon but what about CPU sockets? Intel's roadmap seems pretty set on the LGA1151 socket well into 2019. Who knows when they'll switch on us and then it's a whole new ball game of upgrading. Skylake-X is also on a perpetual course until further notice on the LGA2066 platform. Who know's what Kaby Lake-X is going to be. It's an interesting time for computing. One thing's for sure, though, at some point if you want to upgrade to the latest and greatest from a previous single generation, a system-wide upgrade will be in order. Then again, what's new? But I don't think things have been so obscure for a long while.

I think you may be spending too much time or rather putting too much faith into Youtube reviewer comments section, or just TechPowerUp/Guru3D or some such place in Obvlivion.
Don't infer anything about what Jen-Hsun Huang says, even if it was launching a month from now, there are things he can't say because of the effects it would have on the bottom line and their partner's bottom lines. Or rather, try and go back to any other time since in the last decade where he said anything analogous to "It's coming in a month or Q3 etc?

Intel "struggling" with their 10nm node and the Computex demo have nothing to do with each other (Again, stop reading TPU so much if you are).
1. Those working on Cascade-Lake, Cacade-Lake-X are not even in the same state as those working on Ice-Lake, Cannon-Lake etc. 5GHz demo was not necessary and a complete failure for sure (could have shown 4 GHz with air cooler and still get the record), but platform is coming 100%. in fact INTEL will have 3 platforms for end users, LGA2066 will get 20 and 22 core products, while the "re-purposed" Xeon uses the higher pin count for 24C+ (past 32) cores and six-channel IMC.
You also say re-purpose as if it's a bad thing, SP3 for ThreadRipper is exactly SP3r2 for EPYC, two memory controllers fused off with two sets of PCI-e lanes fused off as well. So anyone can re-purpose, but INTEL shouldn't? BTW, that 24Core demo at Computex was an EPYC platform and CPU.

1a) INTEL is only struggling if you know what their targets are. They have never disclosed this, but somehow we all know just how real the struggle is within INTEL (a company with 100K people mind you and thousands of engineers and physicists spread over two continents at least). Also do you know what they are struggling with and do you know what type of issue their facing? That is, - Is the "struggle" yield, performance, leakage, scaling, a combination or more? This matters because, it can be a problem you work on and solve by throwing more man power into it, or one that is confined to a particular physics phenomenon that needs particular treatment of the die perhaps?

2. There are no scientific limitations, coming anywhere, your grasp of the particular details involved in the silicon business preclude you from making such an assumption. I know enough about it to know just how breathtaking my ignorance is when it comes to process evaluation and IC fabrication. Therefore best not to make assertions from the piercing depths of ignorance, that's how you make sure you stay there.

3. Sockets need only change when there's a technical or financial reason for it. If you didn't know INTEL always asks the board vendors if they want a new socket, and the vendors always say yes. Why? Because that's how they sell new boards. A new CPU on an old socket only means money for INTEL and costs for board vendors at best. HEDT, sockets stay longer because they are (as they have always been) derived from the server platforms and the lifetime there is inherently longer (validation, QC, larger implementations, mission critical work etc).

 

[Mo Facta]

I think you may be spending too much time or rather putting too much faith into Youtube reviewer comments section, or just TechPowerUp/Guru3D or some such place in Obvlivion.
Don't infer anything about what Jen-Hsun Huang says, even if it was launching a month from now, there are things he can't say because of the effects it would have on the bottom line and their partner's bottom lines. Or rather, try and go back to any other time since in the last decade where he said anything analogous to "It's coming in a month or Q3 etc?

Intel "struggling" with their 10nm node and the Computex demo have nothing to do with each other (Again, stop reading TPU so much if you are).
1. Those working on Cascade-Lake, Cacade-Lake-X are not even in the same state as those working on Ice-Lake, Cannon-Lake etc. 5GHz demo was not necessary and a complete failure for sure (could have shown 4 GHz with air cooler and still get the record), but platform is coming 100%. in fact INTEL will have 3 platforms for end users, LGA2066 will get 20 and 22 core products, while the "re-purposed" Xeon uses the higher pin count for 24C+ (past 32) cores and six-channel IMC.
You also say re-purpose as if it's a bad thing, SP3 for ThreadRipper is exactly SP3r2 for EPYC, two memory controllers fused off with two sets of PCI-e lanes fused off as well. So anyone can re-purpose, but INTEL shouldn't? BTW, that 24Core demo at Computex was an EPYC platform and CPU.

1a) INTEL is only struggling if you know what their targets are. They have never disclosed this, but somehow we all know just how real the struggle is within INTEL (a company with 100K people mind you and thousands of engineers and physicists spread over two continents at least). Also do you know what they are struggling with and do you know what type of issue their facing? That is, - Is the "struggle" yield, performance, leakage, scaling, a combination or more? This matters because, it can be a problem you work on and solve by throwing more man power into it, or one that is confined to a particular physics phenomenon that needs particular treatment of the die perhaps?

2. There are no scientific limitations, coming anywhere, your grasp of the particular details involved in the silicon business preclude you from making such an assumption. I know enough about it to know just how breathtaking my ignorance is when it comes to process evaluation and IC fabrication. Therefore best not to make assertions from the piercing depths of ignorance, that's how you make sure you stay there.

3. Sockets need only change when there's a technical or financial reason for it. If you didn't know INTEL always asks the board vendors if they want a new socket, and the vendors always say yes. Why? Because that's how they sell new boards. A new CPU on an old socket only means money for INTEL and costs for board vendors at best. HEDT, sockets stay longer because they are (as they have always been) derived from the server platforms and the lifetime there is inherently longer (validation, QC, larger implementations, mission critical work etc).

Thank you for your very well put together and topic-specific response. I’ll look deeper into what you’re saying and give it more thought. I don’t claim to know everything and I appreciate informed analysis.

Also, I like many others don’t have the hardware on hand to form first-hand opinions via testing so I do rely on what I read and view online. You may be afforded a different situation.

Cheers

 

[Mo Facta]

One more thing. I am not so low-brow as to read YouTube comments and take them as verbatim. Please consider giving me more credit than that.

I avoid that shit show like the plague.

 

[Prosp3ctus]

I think you may be spending too much time or rather putting too much faith into Youtube reviewer comments section, or just TechPowerUp/Guru3D or some such place in Obvlivion.
Don't infer anything about what Jen-Hsun Huang says, even if it was launching a month from now, there are things he can't say because of the effects it would have on the bottom line and their partner's bottom lines. Or rather, try and go back to any other time since in the last decade where he said anything analogous to "It's coming in a month or Q3 etc?

Intel "struggling" with their 10nm node and the Computex demo have nothing to do with each other (Again, stop reading TPU so much if you are).
1. Those working on Cascade-Lake, Cacade-Lake-X are not even in the same state as those working on Ice-Lake, Cannon-Lake etc. 5GHz demo was not necessary and a complete failure for sure (could have shown 4 GHz with air cooler and still get the record), but platform is coming 100%. in fact INTEL will have 3 platforms for end users, LGA2066 will get 20 and 22 core products, while the "re-purposed" Xeon uses the higher pin count for 24C+ (past 32) cores and six-channel IMC.
You also say re-purpose as if it's a bad thing, SP3 for ThreadRipper is exactly SP3r2 for EPYC, two memory controllers fused off with two sets of PCI-e lanes fused off as well. So anyone can re-purpose, but INTEL shouldn't? BTW, that 24Core demo at Computex was an EPYC platform and CPU.

1a) INTEL is only struggling if you know what their targets are. They have never disclosed this, but somehow we all know just how real the struggle is within INTEL (a company with 100K people mind you and thousands of engineers and physicists spread over two continents at least). Also do you know what they are struggling with and do you know what type of issue their facing? That is, - Is the "struggle" yield, performance, leakage, scaling, a combination or more? This matters because, it can be a problem you work on and solve by throwing more man power into it, or one that is confined to a particular physics phenomenon that needs particular treatment of the die perhaps?

2. There are no scientific limitations, coming anywhere, your grasp of the particular details involved in the silicon business preclude you from making such an assumption. I know enough about it to know just how breathtaking my ignorance is when it comes to process evaluation and IC fabrication. Therefore best not to make assertions from the piercing depths of ignorance, that's how you make sure you stay there.

3. Sockets need only change when there's a technical or financial reason for it. If you didn't know INTEL always asks the board vendors if they want a new socket, and the vendors always say yes. Why? Because that's how they sell new boards. A new CPU on an old socket only means money for INTEL and costs for board vendors at best. HEDT, sockets stay longer because they are (as they have always been) derived from the server platforms and the lifetime there is inherently longer (validation, QC, larger implementations, mission critical work etc).

Got any sweet knowledge of whats coming?
I remember you detailing SP3 when it was just a long off rumour and X299 if memory serves.

 

[Gouhan]

Still September

 

[blankster]

Speaking of next gen stuff, I wonder when chips/mobos with PCIe 4.0 support will ever come out... isn't this due or did I miss something?

 

[JollyJamma]

Speaking of next gen stuff, I wonder when chips/mobos with PCIe 4.0 support will ever come out... isn't this due or did I miss something?

Some vloggers have said that it’ll be hear next year but version 5.0 has been on the cards for pretty much a few months after that (6-12 months) and so they may just go right to version 5.

Also, version 3 is only maxed out with a Titan V, overclocked, so I wouldn’t stress. 4 will come but as a formality, I reckon. 5 will then take its place almost immediately.