R9 290 series specifications

[Face2Face]

I'm looking closely to what he is saying for two months or so. Seems he knew details leaked recently a long time ago.

I think sushiwarrior said he had a friend that worked at AMD IIRC.

Bring it on Sushi What else you got for us??

 

[SiliconWars]

Sushi definitely knows some stuff before it's public domain, if you haven't noticed it you haven't been paying close enough attention.

 

[n0x1ous]

Sushi - spill the beans on R9 290 reference pricing

 

[Skurge]

Yeah, Shushi has been spot on on everything. He hasn't given hard numbers, but he said it would be larger than 384bit and it would have more than 2560 shaders, but less than 3072, etc

 

[ShintaiDK]

Because non-Boost GPU clocks don't matter since the card boosts out of the box. So why would I care to link non-boosted clocks? It's just as pointless as linking non-boost GPU clocks for 680/770/780 cards. No one cares. What matters are is minimum boost clocks. For gamers the spec that matters is actually IN-game boost clocks, kinda like GTX770 operating at 1136-1241mhz out of the box, or a reference 780 hitting 990mhz.
http://www.anandtech.com/show/7392/the-geforce-gtx-770-roundup-evga-gigabyte-and-msi-compared/7

Did you remove GTX680's boost clock of 1058mhz too? What's next you are going to tell us to ignore R9 290X's boost clock of 1Ghz, ignore Intel CPU's turbo clocks out of the box and only focus on the non-boost clock? Ya ok.....that is not logical unless you are insinuating false advertising by AMD/Sapphire.

You made a claim that R9 280X will be slower than 7970GE and I showed you that in Sapphire's stack 2 of 3 cards will have faster clocks.



Kinda like after-market 760/770 cards often sell for $0-20 premiums, or kinda like after-market 7950/7970 cards that happen to have rebates, negating the price increase over reference models, implying that there will be R9 280X cards with 7970GE speeds for $300. Kinda like Gigabyte GTX670 Windforce offering boosted clocks, after-market cooler and upgraded VRMs for $0, available for sale on launch day.



And your point is?

R9 280X ~ 7970GE implies for $50 more over 760 2GB, you get a card that's 24% faster at 1080P and 32% faster at 1600P. Game over 760:
http://www.computerbase.de/artikel/grafikkarten/2013/nvidia-geforce-gtx-760-im-test/4/

People on our forum have pointed out for 18 months in a row HD7970/GE's market pricing that undercut 680/770, and yet NV users would say those prices don't matter since we are cherry-picking lowest SKUs with MIRs.

Now R9 280X's official MSRP will be $299, undercutting 770 by $100-150 straight up. That means in 2 months R9 280X with rebates will be hitting $279-289 and making 760-770 completely irrelevant starting next week unless NV drops prices.

Not sure why you try to over dramatise it. And boost speed is not certain, if it was, it wouldnt be called boost speed. So I dont get why you try to compare the 2. Regular R9 280X will not perform like a HD7970GE, it will be slower like a non GE. And that ruins your GTX760 compare as well. Also the 280X is not 100-150$ below the GTX770, its 100$. Its not that hard to calculate. 299 vs 399.

 

[el etro]

Little off topic, but does anyone know when the new Ruby demo will be available for download?

That demo already existed. I don't know if is the same demo but with updated tech this time.

 

[Skurge]

Not sure why you try to over dramatise it. And boost speed is not certain, if it was, it wouldnt be called boost speed. So I dont get why you try to compare the 2. Regular R9 280X will not perform like a HD7970GE, it will be slower like a non GE. And that ruins your GTX760 compare as well. Also the 280X is not 100-150$ below the GTX770, its 100$. Its not that hard to calculate. 299 vs 399.

Making more bold claims? Slower than 7970GE. Tagging this for tomorrow.

 

[ShintaiDK]

Making more bold claims? Slower than 7970GE. Tagging this for tomorrow.

Regular R9 280X 870Mhz baseclock, 1070Mhz boost. HD7970GE, 1000Mhz baseclock, 1050Mhz boost.

 

[OILFIELDTRASH]

Making more bold claims? Slower than 7970GE. Tagging this for tomorrow.

What happens tomorrow?

 

[n0x1ous]

What happens tomorrow?

r9 280X and below (the rebadges) are officialy released

 

[thilanliyan]

Regular R9 280X 870Mhz baseclock, 1070Mhz boost. HD7970GE, 1000Mhz baseclock, 1050Mhz boost.

I can't stand this base clock/boost clock nonsense from nV and AMD. Assuming the specs are the same between the 2, wouldn't the higher boost clock lead to higher performance?

 

[blackened23]

Regular R9 280X 870Mhz baseclock, 1070Mhz boost. HD7970GE, 1000Mhz baseclock, 1050Mhz boost.

1070 boost. Not sure why you're going on and on about it being slower than the GHZ, honestly, because it isn't.

With a 1070mhz boost, that means it is faster than the 7970GE. IIRC the 7970 is always at it's boost clock in 3d games. Unless you're playing a game from 2006 that doesn't require anywhere near what the GPU offers. This isn't dissimilar to the Kepler which is always using the boost clock in 3d games. The Kepler does have a more versatile boost with 1 bin steppings, but the point remains. The boost clock is the only clock that matters.

You can play semantics with the base clock all day long but 3d applications that need the horsepower will use the boost clock of 1070, hence the 280X is indeed faster than the 7970GE.

 

[blackened23]

I can't stand this base clock/boost clock nonsense from nV and AMD. Assuming the specs are the same between the 2, wouldn't the higher boost clock lead to higher performance?

Yes, yes it will. So i'm not sure what the point exactly of arguing over the base clock is. With the Kepler, the base clock is pretty much a worthless data point that everyone ignores and the 280X is the same in that respect. Boost clock is what will be used in 3d games that need the horsepower. Flash games played in a browser window within chrome might use the base clock. And...........who cares. I think the answer would be no one.

What i'm more interested in is whether the VRAM increase on the 280X gives it an appreciable increase over the 7970GE. I think the 680>770 was a decent performance boost, i'm curious as to what the 280X will do in terms of increase over the 7970. My gut tells me not much because it's only a 20mhz boost increase, but i'm not sure. I'm also interested to see if the 280X is voltage locked, and what voltage the 1070mhz boost uses.

 

[SiliconWars]

Regular R9 280X 870Mhz baseclock, 1070Mhz boost. HD7970GE, 1000Mhz baseclock, 1050Mhz boost.

The 280X chips will be higher quality silicon that most GHz Edition chips, so the boost could be the difference in most cases.

 

[smackababy]

Yes, yes it will. So i'm not sure what the point exactly of arguing over the base clock is. With the Kepler, the base clock is pretty much a worthless data point that everyone ignores and the 280X is the same in that respect. Boost clock is what will be used in 3d games that need the horsepower. Flash games played in a browser window within chrome might use the base clock. And...........who cares. I think the answer would be no one.

But then what is the point of the boost clock. Is the GPU not actually capable of running 100% of the time at that clock, so they just allow it during heavy load (temperatures allowing)? Why not market the base clock as being that and have it dynamically throttle "to save energy" or some other marketing BS and run exactly as it does now?

I guess maybe a Turbo mode is more enticing than a Green mode for PC gamers.

 

[blackened23]

No, the boost clock is the only clock speed that ever matters. This is true of both the Kepler and the 7970GHz - only the Boost is used. The base might be used in underwhelming desktop applications such as flash games. But if you're playing Crysis 3 you will only see the boost clock. That's why arguing the base clock is pretty much pointless. It's pointless with both Kepler and Tahiti.

That said, Kepler's GPU Boost is more versatile in that it allows dynamic adjustments based on TDP and temperature, which isn't the case with the 7970GE boost. What generally happens with Kepler cards is that your actual in game boost is far higher than advertised, and if you exceed TDP or temp limits it will dynamically adjust in 13mhz increments to get temp / TDP in line. AMD Boost does not do that. But, even with Kepler you will always be at the boost speed or higher in 3d applications that require the horsepower - the same is true of the 7970GE. Essentially, you will never see base clock in something like Crysis 3 or Metro: LL - you will only see Boost clocks.

Therefore the base clock never matters - Only the boost clock matters in 3d applications. The base clock is only used on the desktop in Flash games, perhaps, as i've mentioned. Personally, I would have liked for AMD to make their version of boost more versatile as is GPU boost 2.0. I've grown to like GPU Boost 2.0 and the concept makes a lot of sense in terms of increasing board longevity. I don't think AMD changed boost at all with the 280X series, but we'll see tomorrow I guess.

 

[ShintaiDK]

I can't stand this base clock/boost clock nonsense from nV and AMD. Assuming the specs are the same between the 2, wouldn't the higher boost clock lead to higher performance?

Yes and no. The boost clock is not a guaranteed clock. If it was, they would have used base clock instead. When the load is light enough, boost works great. When it doesnt work so great you get something like this as the extreme case:

 

[blackened23]

That chart has no bearing on base clock and clockspeeds used in games. It doesn't measure clockspeed at all. If you have a kepler card, if you use software to dynamically watch clockspeeds (such as MSI afterburner), you know what it does. I don't know if you're arguing for arguing' sake because you're wrong. Only boost clock matters. When you surpass TDP or temperature thresholds, your kepler card will dynamically adjust downwards in 13mhz increments. This is what typically happens: let's say you have a GTX 780 with an advertised 1018mhz boost. Your actual boost in 3d games will probably be 1100 or higher. If your temperature passes 80C and you pass TDP limits, you will go down to, let's say, 1083mhz. And it will stay there unless your card has a catastrophic failure where the fans just go kaput. You will always be above boost. I suspect you know this if you have a Kepler card.

The only way for you to go down to base clock is if your fans fail on your card. Other than that you WILL always be at your advertised boost speed or higher, assuming the 3d application in question requires the horsepower. If you're playing a game from 2006 or a flash based game that uses 1% of the GPU power available, then you may be at base clock. But for games that need it, you will always be at boost clockspeeds or higher.

The bottom line is your continued assertion of the 280X being slower than the 7970GHz is incorrect. Period. You have a Kepler card. Do you use EVGA's precision to monitor in game clockspeeds? You know what's up there. You're always boost or higher unless the game is from 2006 or something like that - if the 3d application needs the power, both the 7970GE and Kepler cards will be boost or higher in terms of clockspeeds.

 

[Skurge]

Yes and no. The boost clock is not a guaranteed clock. If it was, they would have used baseclock instead. When the load is light enough, boost works great. When it doesnt work so great you get something like this as the extreme case:

Thats a 7950, we all know 7950 was limited by powertune, which is why it bounces up and down like that. The 7970 was not.

 

[ShintaiDK]

That chart has no bearing on base clock and clockspeeds used in games. It doesn't measure clockspeed at all. If you have a kepler card, if you use software to dynamically watch clockspeeds (such as MSI afterburner), you know what it does. I don't know if you're arguing for arguing' sake because you're wrong. Only boost clock matters. When you surpass TDP or temperature thresholds, your kepler card will dynamically adjust downwards in 13mhz increments. This is what typically happens: let's say you have a GTX 780 with an advertised 1018mhz boost. Your actual boost in 3d games will probably be 1100 or higher. If your temperature passes 80C and you pass TDP limits, you will go down to, let's say, 1083mhz. And it will stay there unless your card has a catastrophic failure where the fans just go kaput. You will always be above boost. I suspect you know this if you have a Kepler card.

The only way for you to go down to base clock is if your fans fail on your card. Other than that you WILL always be at your advertised boost speed or higher, assuming the 3d application in question requires the horsepower. If you're playing a game from 2006 or a flash based game that uses 1% of the GPU power available, then you may be at base clock. But for games that need it, you will always be at boost clockspeeds or higher.

Like this?

Or this?

 

[thilanliyan]

Thank you Intel, AMD, and nVidia for introducing this boost clock nonsense in CPUs and GPUs to muddy the waters.

At least in my case it provides me with no advantages. It just seems like it will start boosting to max in games and scenes that don't require the horsepower anyway...so what's the point of that?

 

[ShintaiDK]

Thats a 7950, we all know 7950 was limited by powertune, which is why it bounces up and down like that. The 7970 was not.

Yes, and AMD says it changes in 4Mhz intervals. Yet all it reports is full boost or no boost. Seems to me AMD knew what a cluster it was and made sure it was just one or the other that got reported.

 

[blackened23]

Like this?

That is EXACTLY WHAT I DESCRIBED EARLIER. 13mhz bin adjustments when you pass temperature or TDP thresholds. You're still above boost clocks even per that chart. You remember what the stock GTX 680 boost clock was, right?
This is what I posted earlier:

When you surpass TDP or temperature thresholds, your kepler card will dynamically adjust downwards in 13mhz increments. This is what typically happens: let's say you have a GTX 780 with an advertised 1018mhz boost. Your actual boost in 3d games will probably be 1100 or higher. If your temperature passes 80C and you pass TDP limits, you will go down to, let's say, 1083mhz. And it will stay there unless your card has a catastrophic failure where the fans just go kaput. You will always be above boost. I suspect you know this if you have a Kepler card.

The GTX 680 stock boost clock is 1058mhz. Unless your card outright FAILS (eg the fans suddenly stop working) it will be above the boost clockspeeds, this applies to both the Kepler and the GHZ 7970. Again, your assertion that the 280X is slower than the 7970GE is wrong. I really feel like you're obfuscating the facts just for arguing sake, to be honest. If you monitor your GPU clockspeeds with EVGA precision with the ingame overlay feature, i'm sure you know what's up.

Aside from this, like I said earlier, AMD's boost is not versatile, unfortunately. I would like for it to be - basically what it does is immediately go to the boost speed when the 3d application requires it. It never dynamically adjusts as the Kepler's boost does. AMD's boost clockspeeds "stick" better than Kepler's, but Kepler has better efficiency as a result of the GPU Boost 2.0 algorithm.

 

[Erenhardt]

I guess 280X will be 7970 with lower voltage, which will lead to lower power draw. You know, those 20 Watts everyone was pointing AMD out, which made 7970 such an inferior product compared to gtx680...

 

[Granseth]

Thank you Intel, AMD, and nVidia for introducing this boost clock nonsense in CPUs and GPUs to muddy the waters.

At least in my case it provides me with no advantages. It just seems like it will start boosting up in games and scenes that don't require the horsepower anyway.

Agreed, it confuses me too.

But as I understand, we can just ignore everything else as performance will be at boost speed (or above) unless the game/program don't need the horsepower.