![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
I am one of the few amd fanboys, so I want to see them succeed over nvidia. One of the main problems is that nvidia has more fanboys and the neutral team is not as large as some people suggest, but to make a play for them you can't leave ANY gaps in the armor for the army of nvidia trolls to spread fud over AMD with.
To that end amd needs to retool to primary things.'
1) Release Cadence - AMD needs new gpu parts across the line every year. Part of the current performance gap is that amd has not really released anything in what seems like almost 2 years. If possible, it would be useful to release an entire lineup of the latest and greatest GCN vs 1.X or whatever it may be called in the futire. From mid range (the low end is gone now) all the way to the top.
This does not mean it has to be on an entirely new architechture, it could just be on a refined process or saving that, a more creatively designed chip. One of the things that's come out about Carrizo is that the chip achieves some of the gains of a die shrink by more densly and intelligently packing the components on the chip. Now that was their apu, they might not have the same room to do that on a gpu if that was already part of the structure, but if gains can be made there, make them.
Efficiency is important as well. Everytime someone makes a chip that takes more power to do more work it seems to work poorly. The pentium 4, the gtx 480, the 290x. The last was a good part but again, the nvidia fanboys will pounce on the competitor doing the same or better work with less energy used. REMOVE those gaps. do not let them have the high ground in those areas or the PRIME Trolls of the world will salivate with glee at the potential to rub salt in those areas and twist the knife they've already lodged in AMDs back when every other post. Hopefully arctic islands renews a focus here, and that will assist amd in getting parts in notebooks as well.
2) Freesync. I wish AMD had a two tier labeling scheme, the more general name currently used would show whether the monitor was tested to support the adaptive sync spec. But AMD needs to go deeper to certify (and reward) monitors makers that go out of their way to create panels with greater vrr windows, like the new nixeus that starts at around 30Hz all the way to 144. Same goes for any overdrive/ghosting issues. These things need to be ironed out to work flawlessly, and I am sure they can be. But when products come onto the market where the freesync alternatives are defficient in some way, even if they are not all freesync displays, it taints the entire picture and casts a pall on the entire lot of them in the minds eye.
PCPER did a hit job on freesync, even before it was tested they threw out presumption after presumption that gsync with their module was inherently superior due to some magical voodoo fairy dust, but you know what? Those smaller vrr windows, coupled with the ghosting issues on that SINGLE monitor, and the behavior of the vrr display below the threshold vs gsync gave them PLENTY of hooks to grapple onto to cheer their own preferred nvidia solutions over AMD while also putting amd down and casting doubt on their entire operation.
This can be solved though. Call it a hyper certification (please someone come up with something cooler than that) where AMD sets out certain properties for a display to hit (i.e. a minimim refresh window and propery synergistic overdrive functions). Those displays will likely command a slight premium due to being designed properly, and AMD gets some cheaper design wins where they slap the smug sneers off the nvidia fanboys looking to troll amd harder.
To that end amd needs to retool to primary things.'
1) Release Cadence - AMD needs new gpu parts across the line every year. Part of the current performance gap is that amd has not really released anything in what seems like almost 2 years. If possible, it would be useful to release an entire lineup of the latest and greatest GCN vs 1.X or whatever it may be called in the futire. From mid range (the low end is gone now) all the way to the top.
This does not mean it has to be on an entirely new architechture, it could just be on a refined process or saving that, a more creatively designed chip. One of the things that's come out about Carrizo is that the chip achieves some of the gains of a die shrink by more densly and intelligently packing the components on the chip. Now that was their apu, they might not have the same room to do that on a gpu if that was already part of the structure, but if gains can be made there, make them.
Efficiency is important as well. Everytime someone makes a chip that takes more power to do more work it seems to work poorly. The pentium 4, the gtx 480, the 290x. The last was a good part but again, the nvidia fanboys will pounce on the competitor doing the same or better work with less energy used. REMOVE those gaps. do not let them have the high ground in those areas or the PRIME Trolls of the world will salivate with glee at the potential to rub salt in those areas and twist the knife they've already lodged in AMDs back when every other post. Hopefully arctic islands renews a focus here, and that will assist amd in getting parts in notebooks as well.
2) Freesync. I wish AMD had a two tier labeling scheme, the more general name currently used would show whether the monitor was tested to support the adaptive sync spec. But AMD needs to go deeper to certify (and reward) monitors makers that go out of their way to create panels with greater vrr windows, like the new nixeus that starts at around 30Hz all the way to 144. Same goes for any overdrive/ghosting issues. These things need to be ironed out to work flawlessly, and I am sure they can be. But when products come onto the market where the freesync alternatives are defficient in some way, even if they are not all freesync displays, it taints the entire picture and casts a pall on the entire lot of them in the minds eye.
PCPER did a hit job on freesync, even before it was tested they threw out presumption after presumption that gsync with their module was inherently superior due to some magical voodoo fairy dust, but you know what? Those smaller vrr windows, coupled with the ghosting issues on that SINGLE monitor, and the behavior of the vrr display below the threshold vs gsync gave them PLENTY of hooks to grapple onto to cheer their own preferred nvidia solutions over AMD while also putting amd down and casting doubt on their entire operation.
This can be solved though. Call it a hyper certification (please someone come up with something cooler than that) where AMD sets out certain properties for a display to hit (i.e. a minimim refresh window and propery synergistic overdrive functions). Those displays will likely command a slight premium due to being designed properly, and AMD gets some cheaper design wins where they slap the smug sneers off the nvidia fanboys looking to troll amd harder.
.
