AMD RX 5700 Navi New arch

[cj09beira]

10 minutes ago, TheDankKoosh said:

ddr means DOUBLE data rate, that doesn't change just because it's gddr. Just because amd calculates it in that way does not mean that it's the true way of calculating it.  

Ps: gddr5x is quad data rate

[TheDankKoosh]

10 minutes ago, cj09beira said:

Ps: gddr5x is quad data rate

Well, that's a small exception, most memory is double data rate. 

[Drak3]

51 minutes ago, Firewrath9 said:

sigh

 

+500mhz on base memory clock is 2500mhz, which is multiplied by 4 to get the effective clock, 10ghz.

Can you do some research on things before saying them?

12 minutes ago, cj09beira said:

Ps: gddr5x is quad data rate

No, GDDR is still a double data rate design. @TheDankKoosh is correct.

 

[RejZoR]

2 hours ago, Drak3 said:

No, GDDR is still a double data rate design. @TheDankKoosh is correct.

 

GDDR5X is QDR. GDDR5 isn't. Though some OC tools measure/display it as such.

[Firewrath9]

4 hours ago, TheDankKoosh said:

ddr means DOUBLE data rate, that doesn't change just because it's gddr. Just because amd calculates it in that way does not mean that it's the true way of calculating it.  

GDDR5X/GDDR6 is 2x GDDR5 is 2x of GDDR3, which is 2x data rate

[Firewrath9]

4 hours ago, Drak3 said:

No, GDDR is still a double data rate design. @TheDankKoosh is correct.

 

its calculated as 4x though in afterburner IIRC

[Drak3]

2 minutes ago, Firewrath9 said:

its calculated as 4x though in afterburner IIRC

And? Software engineers can make mistakes with labeling things.

[Firewrath9]

5 minutes ago, Drak3 said:

And? Software engineers can make mistakes with labeling things.

so if he had +500mhz on mem he has 10ghz actual clock?

Isn't that the arguement?

[Trixanity]

Maybe you guys should just read the spec instead of bickering.

[Drak3]

2 minutes ago, Firewrath9 said:

so if he had +500mhz on mem he has 10ghz actual clock?

Isn't that the arguement?

Don't know, don't really care. I'm not one of the software engineers behind products like Afterburner.

 

But I know this, GDDR5X has what is called quad data rate, but actually changes the prefetch size. To the end consumer, it's the same effect, you're doubling the data transferred relative to clocks.

It's a very misleading naming scheme, and if we apply it to every DDR revision that doubled prefetch sizes, we'd be on hexdecirate/secdecirate (16x rate).

[PlayStation 2]

AMD's STILL using games that no one plays as benchmark material?

[CarlBar]

7 hours ago, Firewrath9 said:

well, first of all, it uses 16Gb (2GB GDDR6) chips, (just 2 of them for every slot). It's basically a RTX Titan with an extra set of 2GB chips and Quadro Drivers.

RTX Titan is 2500$, so at MOST the price of doubling the VRAM is 2500$ (ripping off the GDDR6 on a different titan and putting it on a Titan + Driver = Titan RTX)

Most likely its ~200$, as 12 GB of 8Gb ddr6 is ~50-80$

 

HBM is fine for datacenter usage, but it does not belong on mainstream cards.

Also, engineering problems due to differing heights, which is why several Vegas had TJunction issues

 

 

Interesting, i didn't think GDDR could be setup as 2 chips per "channel", (not sure how else to describe it). I assumed given the 384 bit bus it had to be 12 chips. Explains why i've never heard of 4GB chips. Would love a shot of the PCB to see how they handled the traces on that.

 

Thats said the traces means a 24 chip PCB is going to be really, really expensive. Thats why Consumer cards are limited to 12 chips, the PCB cost of going to more is just unfeasible.To be fair as i implied in my prior post i'm betting a lot of the cost is just the Quadro branding. But i'd bet it's also really freakishly expensive compared to a consumer card, so there's no way they could do that on a consumer level card without majorly inflating the price just like HBM does.

[Totally Average Gameplay]

Let me guess, is NVIDIA gonna counter RX 5700 with an RTX 2070 Ti? They did that with 1070 Ti vs V56?
 

[Humbug]

On 5/31/2019 at 5:15 PM, RejZoR said:

HBM is great actually, but AMD was just ahead of time so much it just wasn't feasible from a supply and price point. You can be assured it'll be a standard sometime in the future like GDDR5/5X/6 is now. Because it's really the only way to go forward as PCB simply cannot carry enough traces to all the memory modules when it comes to large capacity beyond 16GB.

Everybody expected HBM prices to come down, and supply to go up over time.

 

But it didn't happen (so far) and hence AMD was screwed, because they had two successive generations of high end GPUs (Fiji and Vega)  which were designed around HBM and became supply constrained and more expensive to produce than planned.

 

Not sure if that was bad luck or bad planning by AMD.

[cj09beira]

1 hour ago, Humbug said:

Everybody expected HBM prices to come down, and supply to go up over time.

 

But it didn't happen (so far) and hence AMD was screwed, because they had two successive generations of high end GPUs (Fiji and Vega)  which were designed around HBM and became supply constrained and more expensive to produce than planned.

 

Not sure if that was bad luck or bad planning by AMD.

mostly bad luck, problem was those fpgas that launched around the same time were using quite a bit of hbm (and they were expensive so they can afford to pay a premium for it), add the fact that the memory guys were price fixing all memory at the time, and its really a number of unpredictable things happening at the same time

[Trixanity]

1 hour ago, Humbug said:

Everybody expected HBM prices to come down, and supply to go up over time.

 

But it didn't happen (so far) and hence AMD was screwed, because they had two successive generations of high end GPUs (Fiji and Vega)  which were designed around HBM and became supply constrained and more expensive to produce than planned.

 

Not sure if that was bad luck or bad planning by AMD.

Bad planning. It would have been lucky if it actually panned out.

[CarlBar]

2 hours ago, Humbug said:

Not sure if that was bad luck or bad planning by AMD.

 

A bit of both. They really should have had a backup plan ready to go in case GDDR improved enough, but at the same time when they started to move over there was a lot of concern about hitting the limits of GDDR, (NVIDIA used HBM on some of it's last generation professional cards for this reason, they couldn't get the capability they needed out of then available GDDR), However GDDR5X and GDDR6 have extended things further than anyone expected which reduced the need for it which kept prices high. What AMD was gambling on was that NVIDIA would have to switch to HBM for all it's cards as well and that didn't happen, (though as NVIDIA using HBM in a limited fashion last gen shows it was a close thing, and whats going to happen in future is anyone's guess). 

[BiG StroOnZ]

For those that absolutely cannot wait until tomorrow’s official announcement of Radeon RX 5700 series:

Quote

 

 

AMD Radeon RX 5700 XT features 40 Compute Units (2560 stream processors). The Navi GPU is clocked at 1605 MHz in base mode, 1755 MHz in Game Mode and 1905 MHz in boost mode. Of course, the new addition here is the Game Clock.

 

The card is confirmed to feature 8GB of GDDR6 memory. At least two variants will be launching, a 180W TDP (225W TBP) priced around $499 US and a 150W TDP (180W TBP) SKU priced around $399 US.

 

Source:  https://videocardz.com/80966/amd-radeon-rx-5700-xt-picture-and-specs-leaked

[Stefan Payne]

X - doubt.

 

Lets just wait until tomorrow and see what happens...

[RejZoR]

I wonder what they mean with the "game clock". And what does the boost clock serves for then. I mean, in which applications? Or it's guaranteed base clock under heaviest strain, game clock is guaranteed minimum clock for games and boost is the max clock in general that can achieve under good conditions.

[Humbug]

9 hours ago, BiG StroOnZ said:

For those that absolutely cannot wait until tomorrow’s official announcement of Radeon RX 5700 series:

 

Source:  https://videocardz.com/80966/amd-radeon-rx-5700-xt-picture-and-specs-leaked

Look at that hot and loud blower cooler! 

Just what we have all been waiting for.

[ravenshrike]

4 hours ago, Humbug said:

Look at that hot and loud blower cooler! 

Just what we have all been waiting for.

It's not like the usual set don't have production models ready to go. Or did you actually think ASRock's 'concept' coolers were anything but 1-2 slightly tweaked parts away from whatever they ended up putting into production?

[cj09beira]

@Firewrath9

[RejZoR]

Blower coolers blow. I know they improved efficiency, but it won't be quiet no matter what. RX580 and RX590 were cool coz they arrived aftermarket by default.

[cj09beira]

9 minutes ago, RejZoR said:

Blower coolers blow. I know they improved efficiency, but it won't be quiet no matter what. RX580 and RX590 were cool coz they arrived aftermarket by default.

only because rx480 launched first, blowers, would probably be the main thing i would bring up if i ever met with lisa