AMD Ryzen: back in the game?

Pages

Article.

[...]AMD has shown enough of Ryzen to make it an exciting product. For the last five or so years, Intel has remained unchallenged in just about every segment of the PC market, and improvements over time have slowed to a crawl. Its latest chip, the Kaby Lake i7-7700K is an incredibly marginal upgrade over its Skylake predecessor. Even a six-year-old Sandy Bridge i7-2600K can still hold its own when overclocked.

Ryzen might just be the kick in the butt that the market needs—and thus, AMD is now working as hard as possible to convince the sceptical masses that Ryzen is more than just a tech demo, that Ryzen is a real product that will be on shelves before the end of March 2017.

There's plenty room for doubt here: AMD has a history of over-promising and under-delivering (Bulldozer anyone), and AMD was in full control of every demo they showed at CES.

But still: even if they don't fully deliver, it seems like they'll land close to Intel's current line and might finally seduce me into upgrading my 5 year old I2500K.

Thoughts?

Love to see it happen.. but even if its a competitive product AMD has an uphill battle to fight the fanboys that dominate the enthusiast forums. Hopefully the score some much needed OEM contracts and at least get back in that game.

The one thing that really worries me is driver/chipset quality. That's what drove me back to Intel, back in the Athlon 64 days. I ended up spending several hundred dollars on parts for a Linux server that wasn't actually broken, because the BIOS had done something stupid to the AMD chipset, and it was throwing false hard drive errors. I got pissed off and switched it to a tremendously inferior Pentium D.... and it never gave me any more trouble. It was flawless after that. As CPUs, those chips were low quality, but the chipset they were on was excellent. I strongly preferred slower results from a system I could trust.

Speed is great, but the system as a whole needs to work, and keep working. I routinely get a month or more of uptime out of my current machines, sometimes even my desktop. Are the AMD system drivers going to be good enough to offer that?

I am rooting so so so hard for AMD on this release. In particular I want them to *crush* Intel on pricing for roughly equivalent hardware and just force the market to adapt.

I don't know that I believe they will... but I'm going to keep hoping right up until the release.

The Driver issue will continue to plague AMD long after its been resolved. I still hear it all the time in reference to their GPU's which haven't had any more or less issues than Nvidia for at least 7 years if not longer.

Enthusiasts have a LONG memory.. which is ironic in a world of tech that can change so rapidly.

Fingers crossed!

My very first build was an AMD base machine and I LOVED their price points.

That being said, I have been using Intel for a long time now, so I will keep my viewpoints on course with "wait and see".

Enthusiasts have a LONG memory..

Note that my problem was with AMD, not ATI. And this was absolutely endemic to all AMD platforms, that they just flat weren't anywhere near as reliable as Intel.

Intel's dominance is, in many ways, due to their extremely solid chipsets. A machine that you can really trust is worth a lot more than many people think. Everyone looks at the headline speed ratings, but they just take for granted that the computer's not going to crash. That's because we've all gotten used to Intel's excellent engineering. It wasn't something you could count on, back on the Athlon 64.

You're right that my memory is long, when AMD's lousy drivers/BIOS cost me so much in wasted parts.

Did their latest chipsets still have stability issues?

dejanzie wrote:

Did their latest chipsets still have stability issues?

I wondered this, thought about it, then realized I seriously don't know anybody who owns an AMD based PC.

Yeah, that was my thinking, as well. I don't know anyone who runs AMD anymore.

I've built a few 990FX chipset AMD systems for some people in the last 2 years.. had no issues with them under Windows 10.

The Xbox One runs on an AMD APU FWIW and runs a subset of Windows 10 OS.

I only hear horror stories from coworkers who used to work at AMD.

I have had an FX 8320 with a 7870 for 4(?) years now with no problems at all. I've run 8.1 and 10 on it.

I want to upgrade from my overclocked Ivy Bridge i5, but the Intel upgrades available are so marginal (relative to the cost) that it's just not worth it.

Give me something to upgrade to, AMD.

*Legion* wrote:

I want to upgrade from my overclocked Ivy Bridge i5, but the Intel upgrades available are so marginal (relative to the cost) that it's just not worth it.

I'm still on Lynnfield/Nehalem from 2008 (?). As soon as someone can get me a 1 GHz increase without overclocking I'm all in. Not to mention I'd love to actually get a motherboard with modern speeds for SATA and PCIe.

As soon as someone can get me a 1 GHz increase without overclocking I'm all in.

The 7700K does 4.2GHz base, 4.5GHz turbo, without any overclocking at all. With it being a K chip, it might go even a little further, but I probably wouldn't bother with that, myself.

Further, there have been fairly significant IPC improvements since Nehalem; the next gen, Sandy Bridge, was quite a bit better per clock, and then there were marginal improvements in Ivy Bridge and Haswell. (the current Kaby Lake is almost identical to Haswell, clock for clock.) Between the clock speed, IPC, and memory bandwidth improvements, a current machine should have substantially more muscle than what you have. If you're at 3GHz now, I'd guess a 7700K chip would probably be about equivalent to ~5.5GHz for most stuff. For a select few things, it would be a lot faster.

People are pretty regularly hitting 5ghz on 7700k's with a little BIOS twiddling. 4.8 seems reliably and easily achievable.

Still wouldn't upgrade until after Ryzen settles though

Im very excited for the new AMD products. Unfortunately, I just finished my gaming rig. If AMD can hit the price points, they'll skyrocket with popularity. I mean the vega cards are supposed to beat the snot out of the 1080s.

sydaf3x wrote:

I mean the vega cards are supposed to beat the snot out of the 1080s.

Not really the thread for it, but real Ryzen news is slim so let's roll with this temporarily.

I could easily see this being true in well implemented Vulkan/DX12 renderers like Doom, given the way their currently released stuff works, but outside of that I'll believe it when when they actually show something. They're still being super cagey and hiding all their Vega stuff away.

But even if Vega does make a significant leap over the 1080, do you think Nvidia cares at this point? Finally beating the performance of a competing video card basically a full calendar year after it released isn't really an achievement, and Vega's currently revealed release is Q2 this year. That puts it in April at the earliest.

The GTX 1080 is already pushing 10 months old and Nvidia is still holding on to the 1080Ti release. Remember that Ti cards have been as much as a 25% performance increase over the base card in the past. If Vega comes out and is a full 25% jump over the Pascal cards at the high end, but Nvidia instantly matches it with the Ti and cuts pricing on the 1080 on down, then Vega loses a lot of its oomph for anyone other than brand loyalists. And Nvidia can afford to do all of that because, again, they'll have had basically a year riding on one video card lineup with no meaningful competition at the upper segments. The only way AMD saves themselves from this is by winning by *huge* margins on performance at similar pricepoints... or by just beating Nvidia on pricing period, regardless of whatever adjustments they make.

But then loaded up behind the 1080Ti will be preparation for either a full 2xxx series Pascal refresh or the still occasionally rumored early release of the true next gen lineup in Volta, with the move to HBM and everything that brings.

AMD got a good and important win with the RX480, which I have zero qualms recommending over the 1060 at this point, but I'm still skeptical they're doing enough or doing it fast enough to truly push Nvidia off the high end.

Basically the way I see it is Nvidia has to bungle something for AMD to truly end up competitive in the video card market for even most of this year. I hope I'm wrong, but I don't think I am.

I like their chances in the CPU market much better right now, with the big wet fart of a release that is Kaby Lake still lingering.

We don't need AMD to compete with the 1080Ti/Titan anymore (at least not short term) what AMD needs to do is beat Nvidia on the 1060/1070 not just with performance but with price. That's the bulk of the GPU sales both to PC builders and OEM's. Combine that with a price competitive Zen CPU and the possibility of a really strong sub $1000 PC build that outperforms similar $1000 Nvidia/Intel builds is compelling.

TheGameguru wrote:

We don't need AMD to compete with the 1080Ti/Titan anymore (at least not short term) what AMD needs to do is beat Nvidia on the 1060/1070 not just with performance but with price. That's the bulk of the GPU sales both to PC builders and OEM's.

I basically said this though.

The only way AMD saves themselves from this is by winning by *huge* margins on performance at similar pricepoints... or by just beating Nvidia on pricing period, regardless of whatever adjustments they make.

If Nvidia should choose to respond to Vega by releasing the Ti, then just dropping prices across the entire Pascal lineup, literally slotting the 1080 into the 1070's former spot and so on down the line, and then once again put AMD as the subpar option from the midrange on up, what has Vega accomplished for AMD specifically?

It would be good for the consumer market in the short term because hey, more performance for less money. But what does it do for AMD? All it looks like to me is they goosed Nvidia into beating them again.

Whether Nvidia would or would even need to do that, I don't know, but I have to kind of feel like they could if they wanted to, given how long Pascal cards have been out.

And if they were to do it, do you think AMD is as prepared for the rest of this year as Nvidia is?

IDK you seemed focus on the beating the 1080 aspect of Vega. I'm more hopeful that Vega beats the 1070. And for sure Nvidia is holding the 1080Ti back until AMD announces Vega and early benchmarks come out.. they you are all but assured price drops across the entire Nvidia line to make AMD's products less attractive.

It's a losing proposition right now from AMD they are damned if they do damned if they don't. The only hope is if Vega has a leg up architecture wise on Nvidia for the next 3-4 years and in 2018 AMD can put out something compelling that Nvidia cannot easily respond to.

The only viable option for AMD long term seems to be an Intel buyout or partnership which all but dooms the CPU side to the APU market.

I think that AMD will never make headway as long as they focus on one product or market segment. It is way too easy for nvidia or intel to price adjust to be competitive again. Obviously it is good in the short term when they do this but long term stagnation hurts us consumers.
And of course spreading themselves to thin by trying to address too many market segments isn't phi beta kappa either. So I am hoping for a 2-3 pronged approach and it seems obvious that targeting the middle performance range is going to be the most lucrative even if they only capture a moderate portion of that market.

TheGameguru wrote:

IDK you seemed focus on the beating the 1080 aspect of Vega. I'm more hopeful that Vega beats the 1070.

But if they don't even beat the 1080 but focus on beating the 1070... then wouldn't Nvidia shifting all their cards down one slot and dropping prices accordingly after the Ti release hit AMD even harder?

I mean obviously it's all dependent on pricing, but I have trouble believing Vega is going to do what some people seem to think it's going to.

It's a losing proposition right now from AMD they are damned if they do damned if they don't.

I guess I didn't word things as succinctly, but this is the main idea I was getting at. The video card market seems like a no win situation for AMD to me this year barring Vega being just some kind of incredible AMD loyalist wet dream of huge power and amazing pricing. I'd certainly love to see it but... reality.

The only viable option for AMD long term seems to be an Intel buyout or partnership which all but dooms the CPU side to the APU market.

You think? Like even if Ryzen works out and Intel doesn't have some huge new upgrade release they've been sitting on, you think AMD still won't get a foothold in the CPU market?

Kaby Lake is such a crappy boring release I've been feeling like maybe this is AMD's big chance at digging back in on that side.

I just have way less optimism for their GPU stuff.

Personally, I don't see AMD coming back from the brink, for simple financial reasons. They're just too weak now, they've been screwing up way too badly for way too long. Even if both their new products are amazing, Intel and NVidia just have to cut prices and starve them out. Both should be sitting (I haven't actually checked, though) on big piles of the cash they've been raking in by not being challenged in the high end, so all they have to do is accept real tight margins for a year or two, and AMD's probably toast.

The chance of AMD coming up with a CPU that genuinely beats Intel in terms of absolute performance is pretty slim, so they just not going to have pricing power. They can't say, "If you want the fastest CPU, you can have it, for this nosebleed price." Rather, all they can do is match (at best) Intel, who can just cut prices in response.

They have a better chance of delivering something really great with the GPU, but their cards have been mostly about raw power, just throwing lots of transistors at problems. NVidia's have been almost accidentally great, where the tricks they devised for low power usage ended up being real useful, and their cards were able to use relatively cheap architectures to provide surprisingly strong game performance. If AMD really challenges them, they should be able to combine that sleight-of-hand with a bunch more actual transistors.

Right now, NVidia is running fast by figuring out how to avoid doing work they don't have to do. It's that old trick of optimization, the observation that the fastest code of all is code that's not running. They just apply it to transistors instead. They've been able to do a lot with relatively small GPU dies and limited memory bandwidth. If they bulk back up, and go back to saner margins, AMD will need to be doing something far, far more impressive than they've been managing for the last several generations to keep up.

And I just don't think they have the talent, anymore. They've bled too much, for too long, in terms of both engineering talent and money. Even if they manage to deliver perfectly with their reduced skill level, their poor financial position will make it awful easy for their competitors to kill them.

Thin_J wrote:
sydaf3x wrote:

I mean the vega cards are supposed to beat the snot out of the 1080s.

Not really the thread for it, but real Ryzen news is slim so let's roll with this temporarily.

I could easily see this being true in well implemented Vulkan/DX12 renderers like Doom, given the way their currently released stuff works, but outside of that I'll believe it when when they actually show something. They're still being super cagey and hiding all their Vega stuff away.

But even if Vega does make a significant leap over the 1080, do you think Nvidia cares at this point? Finally beating the performance of a competing video card basically a full calendar year after it released isn't really an achievement, and Vega's currently revealed release is Q2 this year. That puts it in April at the earliest.

The GTX 1080 is already pushing 10 months old and Nvidia is still holding on to the 1080Ti release. Remember that Ti cards have been as much as a 25% performance increase over the base card in the past. If Vega comes out and is a full 25% jump over the Pascal cards at the high end, but Nvidia instantly matches it with the Ti and cuts pricing on the 1080 on down, then Vega loses a lot of its oomph for anyone other than brand loyalists. And Nvidia can afford to do all of that because, again, they'll have had basically a year riding on one video card lineup with no meaningful competition at the upper segments. The only way AMD saves themselves from this is by winning by *huge* margins on performance at similar pricepoints... or by just beating Nvidia on pricing period, regardless of whatever adjustments they make.

But then loaded up behind the 1080Ti will be preparation for either a full 2xxx series Pascal refresh or the still occasionally rumored early release of the true next gen lineup in Volta, with the move to HBM and everything that brings.

AMD got a good and important win with the RX480, which I have zero qualms recommending over the 1060 at this point, but I'm still skeptical they're doing enough or doing it fast enough to truly push Nvidia off the high end.

Basically the way I see it is Nvidia has to bungle something for AMD to truly end up competitive in the video card market for even most of this year. I hope I'm wrong, but I don't think I am.

I like their chances in the CPU market much better right now, with the big wet fart of a release that is Kaby Lake still lingering.

You're right. I forgot about the 1080Tis. Im more interested in the price point. If I can get similar performance at a fraction of the cost knowing that catalyst runs great, im all in!

I just bought the RX480 and as time progresses, benchmarks are improving a great deal with dx11 titles.

Thats where AMD prevails. Older cards get better with time when Nvidia seems to drop support for them.

Some supposed price leaks

http://www.forbes.com/sites/antonyle...

While amazing if true.. it still isnt where AMD needs to be.. the amount of 6900K Intel sells vs the more mainstream 7700K and 7600K is where AMD needs a competitive product.

Good, Intel's prices are outrageous.

I mean, if IPC really is as relatively similar as early reports say, they'll likely get me for a 1700X to use in my Server/Encoding PC.

TheGameguru wrote:

While amazing if true.. it still isnt where AMD needs to be.. the amount of 6900K Intel sells vs the more mainstream 7700K and 7600K is where AMD needs a competitive product.

Yeah, but if they can cut the price on a similarly performing 8core/16thread CPU by 70%.... what if the 7700k's 4core/8thread equivalent is $105? Or even $150? The 7700k is sitting around $350 everywhere right now.

Don't get me wrong, I think the lower you get in the lineup the smaller the pricing delta will be, but if they can hit Intel across the board for 30% of the cost, that'd be pretty great.

Ryzen has been tested and runs on Windows 7, but has no official AMD support.

AMD won’t be providing Windows 7 drivers for its upcoming Ryzen processors, the company said, contradicting recent reports that indicated AMD would support Microsoft’s older operating system.

AMD confirmed that it has tested and validated Ryzen on Windows 7, but that it won’t officially support the OS.

Pages