AMD: Volcanic Islands R1100/1200 (8***/9*** series) Speculation/ Rumour Thread

1/4 DP for 290X and 1/2 DP for FirePro - why not doing it now?

Maybe because there are costs associated with it? Just a thought?
Would the 290x selll better if it had resources shifted over to DP, causing it to perform worse relative to nVidias cards? Or alternatively, if it was larger and had even higher power draw and made more noise? AMD could alleviate that by trading die area for power draw, but that in turn would definitely reduce their pricing options and ultimately profitability.

Gamers have no interest in paying for anything really that detracts from gaming performance/$. Having a duopoly makes it possible to force them to do so, but only for as long as your partner/competitor plays along.To extend the DP compute capabilities of the 290, in AMDs competitive market position, would seem ill advised. In most businesses, targeting the priorities of your customers is typically a good idea.
 
In addition, perf/mm doesn't matter at in the professional market.

Performance/watt and heat are king.

IN super computers for example, chips are often passively cooled and such a hot chip would be unusable in such a system.

This is absolute complete and pure bunk. There isn't a supercomputer out there using passively cooled GPUs. Just because a heatsink doesn't have a fan on it doesn't mean it is passively cooled, unless you want to claim that 99.99999999999% of all servers shipped since 1995 are passively cooled as well! All the accelerator cards from AMD, Intel, and Nvidia without a fan have a LFM airspeed requirement. Moving from on card fans to big high speed case fans does not passive cooling make.
 
@3diletante:
One doesn't have to make it more complicated as it is. The number of thermal cycles wouldn't be changed at all by such a priority list. The R290X is specified to run at 1 GHz at 95°C and the default powertune behaviour ensures it reaches said 95°C under load. It can't get that much worse from the durability point of view. Setting a more aggressive fan profile wouldn't make things worse. I'm not against keeping the properties of the rather smooth regulation of the fan speed and such stuff.

===========================

Back to the actual behaviour of Powertune, according to a report in my home forum (3DC), Powertune actually behaves like some kind of a blend between my scenarios (v) and (vi), at least with reduced temp target and increased fan speed limit. It downclocks a bit before it reaches the fan limit, so it keeps the temp target with a combination of downclock and fan speed (with a bit more weight on fan speed apparently). It would be nice if the CCC would provide a switch to decouple the fan control from the temp threshold for downclocking. That way, one could keep the GPU at lower temperatures for "normal" loads and woud only approach 95°C for very demanding stuff (Furmark).

Powertune behaves like this and in this order. The first point is mandatory.
1. Do not supply the GPU with more than 208 watts input

2. try to hit 1 GHz, but
3. do not go over 95 °C and
4. do not go over 40/55 percent PWM impulse

5. if acoustic and thermal limits are hit, reduce clock/voltage until you hit 727 MHz
6. if acoustic and thermal limits are still hit, ignore max PWM impulse
7. if nothing helps, shut down when reaching 100 °C for more than 1 sec.
like said in our review:
http://www.pcgameshardware.de/Gefor...57241/Tests/Radeon-R9-290X-im-Test-1093796/3/

Regarding your suggestion: You can actually give a higher max. Fan speed in the CCC, that is only used if necessary - all depending on your acoustic sensitivity.
 
Anyone see a review where all the voltage bins are listed?
Unless someone finds a way to full dissect the voltage table in the Hawaii BIOS, you won't be seeing that information any time soon. AMD switches between too many states far too quickly, so we can't use logging to find the actual states. What gets reported to logging tools is the average clockspeed over the tick. And on the voltage side there aren't currently any tools (that I'm aware of) that can see the VIDs.
 
I believe that when the new Mac Pros were announced, it was made clear that the GPUs in it would be Tahitis, but I'm not positive on that.

They're FirePro D300's and D500's depending on Mac Pro model, so Pitcairn and Tahiti if I'm not mistaken
 
Regarding your suggestion: You can actually give a higher max. Fan speed in the CCC, that is only used if necessary - all depending on your acoustic sensitivity.
That's all clear, but I still would like the possibility to change the priorities. For instance that it tries to keep the temperature at 85°C, but if that isn't possible at full clock and the maximum allowed fan speed, let the temp rise up to 95°C and only then either increase the fan speed or downclock (whatever the user prefers). A fan speed control (temperature target) decoupled from the temperature downclocking threshold would be nice.
In the end, it really boils down to the question with which priority these three aspects should be obeyed: maximum fan speed, temperature target, clock speed.
 
Option to change priorities would be nice. On top of that, this solution is nowhere near as elegant as Nvidia's Boost 2.0. However, we should be glad that AMD tried to get the most out of each individual ASIC/card. 727mhz for the worst qual asic in warm environments, and 1000mhz for the cherry cold ones. AMD did the opposite... 1000 is the max. Nvidia sets a baseline (~826) and lets the ASIC clock up from there usually winding up stabilizing at it's designated boost clock or into the ~900-950mhz range.

Ironically this increases the 290x's ability to shine on a full cover waterblock @ 40C, because the overclocks will be 50% greater than the throttling clocks while only appearing to be 20% greater than 1000mhz. Even more incentive to go water.
 
That's all clear, but I still would like the possibility to change the priorities. For instance that it tries to keep the temperature at 85°C, but if that isn't possible at full clock and the maximum allowed fan speed, let the temp rise up to 95°C and only then either increase the fan speed or downclock (whatever the user prefers). A fan speed control (temperature target) decoupled from the temperature downclocking threshold would be nice.
In the end, it really boils down to the question with which priority these three aspects should be obeyed: maximum fan speed, temperature target, clock speed.

That would already be a two-stage priority on the temp side. Not sure, if that's possible with the current implementation.
 
What AMD card will be going into the Mac Pro? 290X Firepro variant?

All the design have been made around the Thaiti ones.., there's not yet a FirePro based on hawaii, so even if they are moving to it, this will not be tomorrow, and i suspect they can use it for launch a refresh of the MacPro in the second half of 2014.

Thoses MacPro are not intended to be competitive with other workstations ( PC). Its just for peoples who use allready a MacPro and want to upgrade. Even if the price of those MacPro are really cheap when compared to other workstations. ( I dont know i compare with the type of workstations i use at work for Autocad or Inventor )
 
Last edited by a moderator:
442GB/s of memory throughput! :oops:

With this and the galore of pixel fill-rate, it would be a sin to run this thing on anything lower than a 2160p monitor.
 
Why? Beyond the fact that double is clearly 2x single (only that is sort-of is not). Who cares about this in the desktop / gamer space? And if anybody says "BUT FLIGHTSIMS TOTALLY NEED QUAD-PRECISION FOR COORDINATES" I will cry.

I would argue that 4k displays and true audio are beyond what most enthusiasts need and could afford as well. Amd and nvidia a pitching a range of technologies to justify upgrades to users. GPGPU and DP included. The fact that there is a debate about the DP performance goes to prove that it's just as valid a concern as 4k display performance.
 
I would argue that 4k displays and true audio are beyond what most enthusiasts need and could afford as well. Amd and nvidia a pitching a range of technologies to justify upgrades to users. GPGPU and DP included. The fact that there is a debate about the DP performance goes to prove that it's just as valid a concern as 4k display performance.

4K displays are very expensive, but ultimately "only" have 8,294,400 pixels. A triple 1080p setup has 6,220,800, which isn't far, and quite affordable. A triple 1200p setup has 6,912,000 pixels, which is closer, and still very affordable.

Hell, you can buy 1440p 27" monitors for around $550 (maybe less) and those have 3,686,400 pixels each. Three of them will have 11,059,200 pixels, or quite a bit more than a 4K monitor.

Granted, 3×$550 = $1650 is quite a bit of money, but it's a lot less than a 4K monitor, or even a SLI of Titans/690s.

As for DP, it's for HPC, not for gamers.
 
I would argue that 4k displays and true audio are beyond what most enthusiasts need and could afford as well. Amd and nvidia a pitching a range of technologies to justify upgrades to users. GPGPU and DP included. The fact that there is a debate about the DP performance goes to prove that it's just as valid a concern as 4k display performance.

I agree on the 4K display but I'm not sure about TrueAudio. This is the kind of thing regular PC gamers were gunning for back in the early naughties so I don't see why it would be more than we should expect as standard now.

It's not a marginal improvement like 4K is over 1080p with good AA. The difference between headphones with stereo sound and full positional/reverb etc...should be night and day. I'd say I can't wait except I'm running NV so I won't get to experience this at all :cry:
 
Unless someone finds a way to full dissect the voltage table in the Hawaii BIOS, you won't be seeing that information any time soon. AMD switches between too many states far too quickly, so we can't use logging to find the actual states. What gets reported to logging tools is the average clockspeed over the tick. And on the voltage side there aren't currently any tools (that I'm aware of) that can see the VIDs.

Thanks.
 
Back
Top