• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD to Simultaneously Launch 3rd Gen Ryzen and Unveil Radeon "Navi" This June

Apparently Navi is suppose to be purely designed for Gaming in mind. And not mixing and matching in trying to achieve too much of everything.
This is the key to Navi if sources are correct, and Raja actually fixed Navi properly. Vega had nothing to do with Raja, and was pretty much
targeted to the compute crowd, though it does pretty well in gaming, it probably could have done much much better.

Also speculation based on a couple other sources I remember reading about. Not sure if this is true, but interesting nevertheless. Picture Below:
CW8vxuV2s9Y85OcT.jpg
 
Sounds pretty good, can't wait to see what they offer.
 
I just hope they manage to launch APUs and CPUs together. Them being on separate schedules was confusing in terms of support.
I believe we will see Ryzen 3000 desktop APUs launch along side Ryzen 3000 CPUs as we already have Ryzen mobile 3000 on sale in asus laptops, The laptop and desktop parts are the same except in regards to their TDP and binning. All Ryzen 3000 APUs will be GloFo 12nm, not TSMC 7nm. Vega will be the graphics tech in all ryzen 3000 APUs.
 
I believe we will see Ryzen 3000 desktop APUs launch along side Ryzen 3000 CPUs as we already have Ryzen mobile 3000 on sale in asus laptops, The laptop and desktop parts are the same except in regards to their TDP and binning. All Ryzen 3000 APUs will be GloFo 12nm, not TSMC 7nm. Vega will be the graphics tech in all ryzen 3000 APUs.
For some reason I though Navi would be part of the desktop APUs. The mobile versions are Vega.
 
As an example the first Gen R7 could not go above 4.1 GHZ the 2nd Gen goes to 4.4......so the 3rd gen may do exactly what the rumours have been saying 5.0 GHZ.
I wouldn't count on any huge increases in clock speed. AMD have stated that clock speeds are expected to decline, but I'm not sure if they ment on 7nm or beyond 7nm. Regardless, we have to remember that yields will remain a problem on 7nm until 7nm+ arrives on EUV, so I wouldn't expect the first CPUs on 7nm to achieve the full potential of the node.

If that is the case and they can increase the L1 to L3 cache among other things at 7NM there will be no reason to buy Intel just for gaming anymore.
Intel already achives more performance with less cache. Cache is much more complex than just total size.

While I do expect Zen 2 to close some of the gap in gaming performance, you still have to remember that higher performance per core is still more important for many workloads, especially typical "power users" or workstation workloads. So when Intel (eventually) releases Ice Lake, any IPC changes there will be very relevant for demanding buyers.

This is the key to Navi if sources are correct, and Raja actually fixed Navi properly. Vega had nothing to do with Raja, and was pretty much targeted to the compute crowd, though it does pretty well in gaming, it probably could have done much much better.
Aren't you putting a little too much faith in one guy in management? I don't know about the details about AMD's internals, but it's fairly rare in the industry that people in upper management are deeply involved in actual engineering efforts.
 
Last edited:
Well its all speculation till the actual products are released and benchmarks are conducted.
 
Apparently Navi is suppose to be purely designed for Gaming in mind. And not mixing and matching in trying to achieve too much of everything.
This is the key to Navi if sources are correct, and Raja actually fixed Navi properly. Vega had nothing to do with Raja, and was pretty much
targeted to the compute crowd, though it does pretty well in gaming, it probably could have done much much better.

Also speculation based on a couple other sources I remember reading about. Not sure if this is true, but interesting nevertheless. Picture Below:
View attachment 120161

This thinking is stupid, the Raja Koduri part.

So he worked as head of AMD's GPU side for 5+ years, but he was responsible for NONE of the gpu launches in that time, which was about 4 launches?! lol ok, I keep reading this. Imagine when they hired him they thought 'we won't get anything out of this guy for the first 5 years'.
 
Can't wait to get a Ryzen 3000.
Thank god AMD don't do the "Oh some pin definitions have been changed so old MBs don't work" bull crap.
I had a X370 MB, update the BIOS then change the CPU is just a 5 MINUTES process.

You know they can do what Intel does and just claim the new CPU is incompatible and enforce it in the BIOS.
 
This thinking is stupid, the Raja Koduri part.

So he worked as head of AMD's GPU side for 5+ years, but he was responsible for NONE of the gpu launches in that time, which was about 4 launches?! lol ok, I keep reading this. Imagine when they hired him they thought 'we won't get anything out of this guy for the first 5 years'.

Honestly, there is some truth to it. There are certain points of that process that were established and there was no going back from. Conversely, If Navi was started by Raja, it took years before any tangible product took shape. Think of architects of some of the structures of old that didn't live long enough to see it finished. Not that a GPU and a gothic cathedral are particularly similar but...

It's not like they didn't get anything out of him for a few years.
 
You know they can do what Intel does and just claim the new CPU is incompatible and enforce it in the BIOS.
You know they announced it like 5 months ago X370 supports Ryzen 3000 Series just by updating the BIOS.
And the new BIOS is released mid March.
 
I'm...a smartass.

Ok <3

Also what do we all think, ps5 for this Xmas with the cream of the crop from the early batches?
 
This thinking is stupid, the Raja Koduri part.

So he worked as head of AMD's GPU side for 5+ years, but he was responsible for NONE of the gpu launches in that time, which was about 4 launches?! lol ok, I keep reading this. Imagine when they hired him they thought 'we won't get anything out of this guy for the first 5 years'.

Here was my point about Raja, Many seem to believe VEGA was already designed as it is and that nothing Raja could have done to boost it higher for gaming performance. Something of that matter, and that Raja helped build Navi?? I don't know, there's so many conflicting reports. There is one thing I know for a FACT, AMD has kept NAVI very tight lipped LOL

AMD’s Navi GPU, the first non-GCN GPU, is ready for launch
AMD Navi - None GCN based....
 
I still think that it's integrated Navi, not discreet Navi. Remember, Sony needs an APU for PS5 which likely fast tracked Zen 2 and integrated Navi development. Sony needs to get prototype consoles out to developers now so they can get games finished for a Holiday 2020 release.

With the APUs out of the way, they can focus on making discreet Navi chips for AIBs for a Holiday 2019 release.
 
OK, I'll be Machiavellian type and play devil's advocate for a sec. :cool:

What is there after Zen for AMD? Yes it has its own 5 minutes right now. Intel is in the doldrums. Since first iteration Zen platform matured a lot, lot, lot. But route which AMD pursues (and Intel) is a finite one. Yes servers may benefit from more cores (up to a point only!), but anything above 12-16 cores for casual users (gamers in particular) is total madness.

Even from *cough* Pro *cough* stand-point. CPU is not really needed to do things at this moment in time. GPUs offer vastly superior computing power across the board. Simple test between rendering on IRAY done this week, with CPU assist on 12 core or 32 core TR, makes less than minute or just tad over 1 minute of difference on a nearly 31h render vs GPU alone (single 1080Ti, finished in 30.47h exactly). I won't even attempt that render on CPU alone, should take rest of the year... CPU uselessness in a nutshell.
 
OK, I'll be Machiavellian type and play devil's advocate for a sec. :cool:

What is there after Zen for AMD? Yes it has its own 5 minutes right now. Intel is in the doldrums. Since first iteration Zen platform matured a lot, lot, lot. But route which AMD pursues (and Intel) is a finite one. Yes servers may benefit from more cores (up to a point only!), but anything above 12-16 cores for casual users (gamers in particular) is total madness.

Even from *cough* Pro *cough* stand-point. CPU is not really needed to do things at this moment in time. GPUs offer vastly superior computing power across the board. Simple test between rendering on IRAY done this week, with CPU assist on 12 core or 32 core TR, makes less than minute or just tad over 1 minute of difference on a nearly 31h render vs GPU alone (single 1080Ti, finished in 30.47h exactly). I won't even attempt that render on CPU alone, should take rest of the year... CPU uselessness in a nutshell.

Amd innovates and then intel duplicates. It is how the CPU industry has been for the past 20 years.

By the time intel has their own Zen style CPU out and leap frogs amd they will be looking past silicon to the next tech as we are in the twilight of this technology's lifespan.

Games are designed to make use of 4 core and 8 threads as that is what hardware the majority of gamers have had. As the hardware increases so will softwares demand for it.

We haven't really had another crysis moment for 12 years as it didn't sell. Nobody could run it so nobody bought it. Now we have hardware waiting for software to utilise it, but it will get used. Trust me :)
 
Amd innovates and then intel duplicates. It is how the CPU industry has been for the past 20 years.

Care to explain what Bulldozer through Excavator brought to the CPU industry other than disappointment?
 
Care to explain what Bulldozer through Excavator brought to the CPU industry other than disappointment?
Nothing. I think you misunderstand his factual point. Everything before Bulldozer, AMD innovated and Intel followed. Yes the gorilla followed the monkey, this is a well known industry fact that's easily researchable online.

AMD fell behind with Bulldozer, an obvious engineering and design mistake. That was because they relied too much on automation. Jim Keller was hired and they designed ZEN. The story continues........

Intel took advantage of AMDs Bulldozer and gained market share.......
 
Nothing. I think you misunderstand his factual point. Everything before Bulldozer, AMD innovated and Intel followed. Yes the gorilla followed the monkey, this is a well known industry fact that's easily researchable online.

AMD fell behind with Bulldozer, an obvious engineering and design mistake. That was because they relied too much on automation. Jim Keller was hired and they designed ZEN. The story continues........

Intel took advantage of AMDs Bulldozer and gained market share.......

AMD gave us 64 bit and Intel gave us hyper threading. There really hasn't been anything else in 20 years that changes the game for the average Joe. Clock speeds and core counts don't really make the list as game changers.
 
So multicores don't count, really or even IMC :rolleyes:

I mean IBM had the first multi core processor, and 64 bit. So what are you getting at?
 
what a time to be alive... 7nm Navi & 3rd Gen Ryzen releasing at the same time.

I can't wait to see what comes of this announcement.
I have some bucks put away and I'm ready to pounce if reviews are good.
 
I mean IBM had the first multi core processor, and 64 bit. So what are you getting at?
And IBM also had CMT, the first SMT processor wasn't even from Intel. But we are talking about consumer chips presumably btw what about IMC, which you ignored?
 
And IBM also had CMT, the first SMT processor wasn't even from Intel. But we are talking about consumer chips presumably btw what about IMC, which you ignored?

Touche. As for IMC, what did it do for us? Did it significantly drive up performance? Did it significantly drive down power consumption? Or was it more akin to adding a 6th gear to a 5 gear transmission?

I am referring to innovations that actually matter to end users. Packing 12 cores into a cpu really isn't all that impressive when 6 faster cores is objectively better (of course, there are use cases for 12). But those 12 core CPUs really aren't earth shattering...it's just meh.

Edit 2: Also, I'm not AMD hating before we get there. Ryzen 3000 series is going to be my upgrade path (unless it totally bombs or is like unobtanium).
 
The biggest advantage was mem latency, but with Zen2 we're likely going backwards in that direction. There's also (slight) power savings which could also be lost with this latest chiplet approach. If Physics wasn't such a big downer I think AMD wouldn't have had to move forward with this approach. Definitely more like another gear, however that one change drove performance a big way in the last ~1.5 decades.
 
Back
Top