** The Official Nvidia GeForce 'Pascal' Thread - for general gossip and discussions **

[Pottsey]

Not really
http://www.anandtech.com/show/735/10

That shows it was more powerful then the competition it was against. Kyro II has a low end card with lower raw specs then a GeForce MX, was priced against a GeForce MX but as you can see in those benchmarks was double the speed right up to being only 11% below a GeForce Pro at 1600x1200x32. It wasn’t the best of the best but its price performance was amazing at the time. That link really shows off the efficiency advantage PowerVR have over AMD or NVidia.

Check out the FSAA benchmarks. http://www.anandtech.com/show/735/16 right past the GeForce Ultra. Not bad for a low end card.

 

[D.P.]

Its price was set because of the performance, if it performed like a GF2 Pro then it would be priced at that level.

 

[Mauller]

That shows it was more powerful then the competition it was against.

Look at the other slides and compare. You chose one instance out of all of the benchmarks. Cherry-picking does not make it better overall.

It would have been nice if 3DFX didn't go under and if Imagion remained in the Desktop GPU business. but that never happened.

 

[Pottsey]

Its price was set because of the performance, if it performed like a GF2 Pro then it would be priced at that level.

That makes no sense as it performance was often high end but its price was low end. It did at times out perform a GeForce Pro.

The reason it was priced low end was because its specs are low end. The Kyro 2 was SDR ram doing 350 MOperations/s and a fillrate of 350. While the GeForce 2 GTS was DDR ram with 800 MOperations/s and a fillrate of 800.

Kyro was under half the raw specs but keeping pace in speed. Performance per watt and performance per price was amazing at the time. It wasn't the best at everything but for the price it was a interesting chip.

 

[Jono8]

ignore

 

[FoxEye]

Well one thing's for sure, it's a real shame there were no desktop PC cards after the Kyro II. I too had one, and it was a cracking little card.

It's almost mystifying that nobody made PowerVR cards for the PC after that point.

 

[D.P.]

That makes no sense as it performance was often high end but its price was low end. It did at times out perform a GeForce Pro.

The reason it was priced low end was because its specs are low end. The Kyro 2 was SDR ram doing 350 MOperations/s and a fillrate of 350. While the GeForce 2 GTS was DDR ram with 800 MOperations/s and a fillrate of 800.

Kyro was under half the raw specs but keeping pace in speed. Performance per watt and performance per price was amazing at the time. It wasn't the best at everything but for the price it was a interesting chip.

No compnay sells products based on costs, they sell products based on market value. Teh Kyro simply didn't have the value of the alternative due to its performance.

It did not frequently keep up with the competitors, which is why they didn't sell well, and why PowerVR didn't release another GPU. If PowerVR made competitive cards they would have seen good sales and continued to operate in the market but they failed.

 

[FoxEye]

No compnay sells products based on costs, they sell products based on market value. Teh Kyro simply didn't have the value of the alternative due to its performance.

It did not frequently keep up with the competitors, which is why they didn't sell well, and why PowerVR didn't release another GPU. If PowerVR made competitive cards they would have seen good sales and continued to operate in the market but they failed.

That's just not true.

They were well reviewed on release, and in games that didn't need hardware T&L they were awesome. I bought one on the strength of very positive magazine reviews at the time (was it PC Format or PC Zone, I don't remember).

It was the lack of hardware T&L that rendered them obsolete, but on release they performed fantastically well in everything else.

To say it wasn't competitive is just wrong.

 

[Pottsey]

No compnay sells products based on costs, they sell products based on market value. Teh Kyro simply didn't have the value of the alternative due to its performance.

I don't understand your point of view. You post evidence they did perform well in key games of the time then say they didn't sell for much value due to low performance. Anyway performance had nothing to do with the price or reason for the price.

It did not frequently keep up with the competitors, which is why they didn't sell well, and why PowerVR didn't release another GPU. If PowerVR made competitive cards they would have seen good sales and continued to operate in the market but they failed.

They did sell well in the millions in fact. Not keeping up with the competitors and unit volume had nothing to do with why there wasn't more cards. It wasn't PowerVR that pulled out of the market. But this isn't the thread to make longs posts on PowerVR history.

 

[CAT-THE-FIFTH]

That's just not true.

They were well reviewed on release, and in games that didn't need hardware T&L they were awesome. I bought one on the strength of very positive magazine reviews at the time (was it PC Format or PC Zone, I don't remember).

It was the lack of hardware T&L that rendered them obsolete, but on release they performed fantastically well in everything else.

To say it wasn't competitive is just wrong.

Yep,agreed and they did far better than equivalently priced cards from both Nvidia and ATI at the time:

http://www.anandtech.com/show/735

We are all extremely lucky that Kyro II based products will not only be available, but also easy to get by the end of March or the beginning of April. The performance of the Kyro II based 3D Prophet 4500 is nothing short of stunning given its price: a mere $149.99.

It has been a while since we have had a truly high powered graphics card dip below the $200 price mark. In the past, stripped-down versions of higher performance parts were sold to cost-conscious consumers, oftentimes leaving them with sub-par performance. The Kyro II changes all that.

With its tile based rendering algorithm, the Kyro II provides blazing fast performance considering the price and was actually able to beat products almost $200 more than the cost of a Kyro II based board. Throughout the benchmarks, the Kyro II based 3D Prophet 4500 simply dominated everything else in its price range. The Kyro II was ready and able to tackle any game we sent its way.

 

[Boomstick777]

GDDR5X is looking promising, I would prefer HBM 2.0 obviously but looks like it's not possible for the entire range at first.

The GDDR5X standard brings three key improvements to the well-established GDDR5: it increases data-rates by up to a factor of two, it improves energy efficiency of high-end memory, and it defines new capacities of memory chips to enable denser memory configurations of add-in graphics boards or other devices.

even a 256-bit GDDR5X memory sub-systems running at 14 Gbps could provide up to 448 GBps of memory bandwidth, just 12.5% lower compared to that of AMD’s Radeon R9 Fury X

http://cyberparse.co.uk/2016/01/22/...d-by-jedec-new-graphics-memory-up-to-14-gbps/

 

[CAT-THE-FIFTH]

GDDR5X is looking promising, I would prefer HBM 2.0 obviously but looks like it's not possible for the entire range at first.





http://cyberparse.co.uk/2016/01/22/...d-by-jedec-new-graphics-memory-up-to-14-gbps/

So would that mean the lower end Polaris card with a 128 bit memory controller have the same bandwidth as a R9 380X??

 

[Mauller]

So would that mean the lower end Polaris card with a 128 bit memory controller have the same bandwidth as a R9 380X??

Would be dependent on quality and clock speed of the memory. But a bus that is too thin can also add latency in some workloads.

 

[Orangey]

It won't reach the top speeds until 2018 or later. It's on the graph.

 

[drunkenmaster]

So would that mean the lower end Polaris card with a 128 bit memory controller have the same bandwidth as a R9 380X??

No, double speed gddr5x won't be available in the first wave, much like HBM2 wasn't available nor GDDR5 8Gbps chips at first. It looks like it will be between 10-12Gbps first chips made available, so up to 50% faster.

Power and capacity are rather up in the air with most of these sites being pretty disingenuous, only 8Gb capacities appear to be coming in the first chips which is also available in GDDR5. By the time capacity and speed of GDDR5x ramp up(probably over 2-3 year span) HBM2 (and beyond) will be in very large scale production with interposers and packing of products using them extremely widely used. So GDDR5x looks to be a fairly short lived usefulness. Power wise these slides keep saying gddr5 is limited to 1.5v thus 1.35v gddr5x is lower power.... when we have Samsung and Hynix both shipping gddr5 chips using 1.35v.

I would actually expect say 20nm gddr5x 12Gbps to use noticeably more power than a 20nm gddr5 7Gbps chip. The question is how much, enough more that lower bus and slower memory uses less power or not. It still might be gddr5 on the low end, gddr5x used on midrange where for instance something a bit faster than Fury X might want 400+ Gbps and you can't do it with gddr5 on a mid sized die. Then HBM2 being reserved for the high end.

I've seen no graphs comparing gddr5x to gddr5 for power per pin or power per GB/s as we got with HBM vs GDDR5.

 

[Gregster]

I am excited about any improvement in memory technology that will increase GPU performance, but I can tell you that from my conversations with both AMD and NVIDIA, no one appears to be jumping at the chance to integrate GDDR5X into upcoming graphics cards. That doesn't mean it won't happen with some version of Polaris or Pascal, but it seems that there may be concerns other than bandwidth that keep it from taking hold.

http://www.pcper.com/news/Graphics-Cards/GDDR5X-Memory-Standard-Gets-Official-JEDEC

 

[Mauller]

Unless Hynix were in touch with AMD and Nvidia early on and considering it was only just made a standard. then their memory controller designs would not work with GDDR5X since it requires new hardware to work.

So in the end it is unsurprising that the first batches of Polaris and Pascal may not use GDDR5X. Also if you consider the scale of manufacturing then only hynix will have chips.

 

[Simon Belmont]

Unless Hynix were in touch with AMD and Nvidia early on and considering it was only just made a standard. then their memory controller designs would not work with GDDR5X since it requires new hardware to work.

So in the end it is unsurprising that the first batches of Polaris and Pascal may not use GDDR5X. Also if you consider the scale of manufacturing then only hynix will have chips.

So Nvidia and AMD told you all there plans on there new cards?

Until Nvidia and AMD say anything all this is a dream. GTC should bring real news not fantasy.

 

[Simon Belmont]

Nvidia Pascal/HBM2 GPUs rumoured to launch in 2nd half of 2016

http://www.kitguru.net/components/g...-gpus-rumoured-to-launch-in-2nd-half-of-2016/

 

[BriT]

so 5 months away?