Shootout: 780 Lightning vs 290

[blackened23]

Linus tech tips did a reference 290 under water versus GTX 780 reference under water:

https://www.youtube.com/watch?v=RqaHh-y51us&feature=*********&a

GTX 780 reference: 4 phase PCB, GPU boost 2.0, 1.2V voltage limit
R9-290 reference: 6 phase PCB with AMD's boost, 1.25V voltage limit (?")

Net gain: GTX 780 gained 9% at maximum overclocks
Net gain: R9 -290 gained 7% at maximum water overclocks

GTX 780 @ 35C plus/minus a few at 100% GPU load. R9-290 @ ~45C+/- a few degrees @ 100% GPU load.. Gear used:

Koolance Gear
R9 290 Block: Koolance
GTX 780 Block: Koolance
Quick Disconnect Fittings: Koolance
Tube Reservoirs: Koolance
DDC Pump Top: Koolance

Maximum overclock with 24/7 stability in all games - core clock actual: GTX 780 ~1180MHz, R9-290 1110MHz under water per clock listing master spreadsheet:

https://docs.google.com/spreadsheet...by1jNTFvaTF2UHhaMzdMQ0FNM2c&usp=sharing#gid=0

 

[BallaTheFeared]

Linus tech tips did a reference 290 under water versus GTX 780 reference under water:

https://www.youtube.com/watch?v=RqaHh-y51us&feature=*********&a

GTX 780 reference: 4 phase PCB, GPU boost 2.0, 1.2V voltage limit
R9-290 reference: 6 phase PCB with AMD's boost, 1.25V voltage limit (?")

Net gain: GTX 780 gained 9% at maximum overclocks
Net gain: R9 -290 gained 7% at maximum water overclocks

Both GPUs were at 35-45C at 100% GPU load. Gear used:

Koolance Gear
R9 290 Block: Koolance
GTX 780 Block: Koolance
Quick Disconnect Fittings: Koolance
Tube Reservoirs: Koolance
DDC Pump Top: Koolance

Maximum overclock with 24/7 stability in all games - core clock actual: GTX 780 ~1180MHz, R9-290 1110MHz under water per clock listing master spreadsheet:

https://docs.google.com/spreadsheet...by1jNTFvaTF2UHhaMzdMQ0FNM2c&usp=sharing#gid=0

Shocking!

No, no actually it's not it's been clear since the first WC results and OC results starting pouring in off the internet. :whiste:

 

[blackened23]

Well the big difference here is the 290 is voltage unlocked (from what i've read and from what Elfear indicated) while the GTX 780 with a reference PCB is not. Not unless you do some trickery, which linus did not use a skynet BIOS or the afterburner hack.

 

[Silverforce11]

Shocking!

No, no actually it's not it's been clear since the first WC results and OC results starting pouring in off the internet. :whiste:

Actually the first OC results was with the reference fan, Overclockers and Gibbo did an OC chart, R290 @ 1.2ghz. There seems to be no major difference on air or water, silicon lottery applies but the limit seems to be around 1.25ghz for R290/X, haven't seen many users report higher than that.

 

[Fastx]

Hey guys some quick questions, when I read at HardwareCancucks on the EVGA 780 SC ACX 967/1020 he list the "boost average clock speed" at 1123. So I am taking this as out of the box without over clocking of course and this would be the in gaming boost clock right?

Also how much can the above boost average clock speed of 1123 from the exact same card above vary from card to card?

I plan to overclock on stock voltage and hope to hit around 1210-1240 in gaming boost I think there is a decent chance with these cards (EVGA 780 SC ACX) from review sites I have read. But if not I don't mind having to go just a little higher on the voltage if needed like to 1.21-1.22 but I think I am limited to 1.2 with PX/AB. With that said what is the max voltage I could with the Precision X software or AB using the V offset?

So not looking to do a bios update to get higher voltage at this time and I take the classified voltage controller to raise the voltage would require a bios change for it to work correct?

Just waiting for my card which should be here Saturday so just gathering info on the 780 till then.

 

[jj109]

Hey guys some quick questions, when I read at HardwareCancucks on the EVGA 780 SC ACX 967/1020 he list the "boost average clock speed" at 1123. So I am taking this as out of the box without over clocking of course and this would be the in gaming boost clock right?

Also how much can the above boost average clock speed of 1123 from the exact same card vary from card to card?

I plan to overclock on stock voltage and hope to hit around 1210-1240 in gaming boost I think there is a decent chance with these cards from review sites I have read. But if not I don't mind having to go just a little higher on the voltage if needed like to 1.21-1.22 but I think I am limited to 1.2 with PX/AB. With that said what is the max voltage I could with the Precision X software or AB using the V offset?

So not looking to do a bios update to get higher voltage at this time and I take the classified voltage controller to raise the voltage would require a bios change for it to work correct?

Just waiting for my card which should be here Saturday and so gathering info on the 780 till then.

I've only used two different GTX 780s and never ran them stock so I can't tell you their stock boost clocks, but they don't seem to vary more than a couple bins (so a small multiple of 13 MHz).

Both 780s I owned hit 1230 MHz boost with default maximum boost voltage (1.2 V), but that varies from card to card. To get 1.212 V, you need to use a custom vBios that unlocks the boost bin. Unlocking it will also make the card boost higher, sometimes into instability. I find that best results are with SkyN3t vBios that unlocks the last bin and disables boost so you end up with more control over the core clock. The current 780 I use has unlocked voltage, so it can hit 1350 MHz with 1.28 V. I don't need it and the voltage hack locks the voltage to 1.28 V and prevents the card from going into low power, so I just use the 1.212 V max.

 

[ICDP]

Linus tech tips did a reference 290 under water versus GTX 780 reference under water:

https://www.youtube.com/watch?v=RqaHh-y51us&feature=*********&a

GTX 780 reference: 4 phase PCB, GPU boost 2.0, 1.2V voltage limit
R9-290 reference: 6 phase PCB with AMD's boost, 1.25V voltage limit (?")

Net gain: GTX 780 gained 9% at maximum overclocks
Net gain: R9 -290 gained 7% at maximum water overclocks

GTX 780 @ 35C plus/minus a few at 100% GPU load. R9-290 @ ~45C+/- a few degrees @ 100% GPU load.. Gear used:

Koolance Gear
R9 290 Block: Koolance
GTX 780 Block: Koolance
Quick Disconnect Fittings: Koolance
Tube Reservoirs: Koolance
DDC Pump Top: Koolance

Maximum overclock with 24/7 stability in all games - core clock actual: GTX 780 ~1180MHz, R9-290 1110MHz under water per clock listing master spreadsheet:

https://docs.google.com/spreadsheet...by1jNTFvaTF2UHhaMzdMQ0FNM2c&usp=sharing#gid=0

The GTX 780 was only using +38mV and the R9 290X was not using any extra volts at all in these tests. 1180 an a WC GTX780 and 1110 on a WC R9 290X is very low.

 

[Silverforce11]

The GTX 780 was only using +38mV and the R9 290X was not using any extra volts at all in these tests. 1180 an a WC GTX780 and 1110 on a WC R9 290X is very low.

What exactly is the point of doing a watercooling OC test when they OC on stock volts... /fail

 

[ICDP]

I've only used two different GTX 780s and never ran them stock so I can't tell you their stock boost clocks, but they don't seem to vary more than a couple bins (so a small multiple of 13 MHz).

Both 780s I owned hit 1230 MHz boost with default maximum boost voltage (1.2 V), but that varies from card to card. To get 1.212 V, you need to use a custom vBios that unlocks the boost bin. Unlocking it will also make the card boost higher, sometimes into instability. I find that best results are with SkyN3t vBios that unlocks the last bin and disables boost so you end up with more control over the core clock. The current 780 I use has unlocked voltage, so it can hit 1350 MHz with 1.28 V. I don't need it and the voltage hack locks the voltage to 1.28 V and prevents the card from going into low power, so I just use the 1.212 V max.

I owned 2x MSI TFII GTX 780 Gaming cards and in order to get over 1200 core I need to use the voltage modded BIOS. I also tested one with 1280 core at 1.28v but felt that as they were on air this was not sustainable.

IMHO people should expect ~1180 with stock BIOS and ~1220-1240 with the modded BIOS. Anything over this will be seen as a bonus.

 

[Leadbox]

Hey guys some quick questions, when I read at HardwareCancucks on the EVGA 780 SC ACX 967/1020 he list the "boost average clock speed" at 1123. So I am taking this as out of the box without over clocking of course and this would be the in gaming boost clock right?

Also how much can the above boost average clock speed of 1123 from the exact same card above vary from card to card?

I plan to overclock on stock voltage and hope to hit around 1210-1240 in gaming boost I think there is a decent chance with these cards (EVGA 780 SC ACX) from review sites I have read. But if not I don't mind having to go just a little higher on the voltage if needed like to 1.21-1.22 but I think I am limited to 1.2 with PX/AB. With that said what is the max voltage I could with the Precision X software or AB using the V offset?

So not looking to do a bios update to get higher voltage at this time and I take the classified voltage controller to raise the voltage would require a bios change for it to work correct?

Just waiting for my card which should be here Saturday so just gathering info on the 780 till then.

I wouldn't worry too much about high it boosts, chances are when you see those really high boost clocks nothing much is happening. Boost loves to stretch out on these nothing moments

 

[ICDP]

What exactly is the point of doing a watercooling OC test when they OC on stock volts... /fail

My thoughts exactly. I would be testing with the specially modded BIOS aimed at people with watercooled cards that allow much higher voltage than stock BIOS.

 

[blackened23]

The GTX 780 was only using +38mV and the R9 290X was not using any extra volts at all in these tests. 1180 an a WC GTX780 and 1110 on a WC R9 290X is very low.

[citation required]

Did you seriously just make that up or what. Because nowhere did he say that. Because I see +25% power limit on the R9-290, with 106% power limit and temp target 95C on the GTX 780 as mentioned in the video and in the google document.

I'd say just like all of the evidence so far leads us to believe, the 290 and 290X cards just don't have the overclocking headroom or scaling that GK110 has even under water.

All published data so far indicates:

Reference GK110 has more overclocking headroom and scaling than the reference 290/X.
Aftermarket GK110 has more overclocking headroom and scaling than aftermarket 290/X (as seen with all DC II 290X reviews)
Reference GK110 has more overclocking headroom and scaling under water than the reference 290/X, despite having fewer power phases.

Oh yeah. "Just you wait" for this and this card in March of next year. That will really overclock.

 

[Silverforce11]

[citation required]

Did you seriously just make that up or what. Because nowhere did he say that. Because I see +25% power limit on the R9-290, with 106% power limit and temp target 95C on the GTX 780 as mentioned in the video and in the google document.

I'd say just like all of the evidence so far leads us to believe, the 290 and 290X cards just don't have the overclocking headroom or scaling that GK110 has even under water. Oh yeah I forget. "Just you wait" for this and this card in March of next year. That will really overclock.

Its in the document you linked...

Look under core voltage. 780 WC = 38, R290 = null.

 

[ICDP]

[citation required]

Did you seriously just make that up or what. Because nowhere did he say that. Because I see +25% power limit on the R9-290, with 106% power limit and temp target 95C on the GTX 780 as mentioned in the video and in the google document.

I'd say just like all of the evidence so far leads us to believe, the 290 and 290X cards just don't have the overclocking headroom or scaling that GK110 has even under water. Oh yeah I forget. "Just you wait" for this and this card in March of next year. That will really overclock.

Oh my! Did you read the link you thankfully provided? +25% power limit and +106% power limit does not = voltage increase.

The spreadsheet shows very clearly that the GTX780 had +38mV added and the voltage and columns for the R9 290 is actually blank. They didn't touch voltage on the R9 290 and they used stock voltage increase on the GTX780.

It seems your latest agenda is to prove the R9 290/X is less of an overclocker than GK110, as if that suddenly renders R9 290/X cards as inferior. Even if that were an absolute fact (it isn't as we all know silicone lottery applies) it does not change the fact that these cards are so close in performance we finally have competition and reduced prices. Now before you go off about mining causing price gouging on AMD cards that was not the case at release. Remember when Nvidia felt R9 290X (let alone R9 290) was such a threat they reduced GTX780 prices by $150 - $200?

R9 290 ~ GTX780
R9 290X ~ Titan/GTX780Ti

Giver or take a few % all of the above cards have similar performance and we the consumers finally have competition forcing prices down.

Instead of bitching about the "crap reference coolers", "True Audio meh", "not as much OC potential", "VRM temperatures", or (insert reason why R9 290/X sucks here). Why not just accept that the R9 290/X are actually bloody good GPUs that give AMD parity with Nvidia? Sometimes I wonder if people preferred it when Nvidia could charge extortionate prices due to lack of competition.

 

[krumme]

Oh my! Did you read the link provided to the google docs? +25% power limit and +106% power limit does not = voltage increase.

The spreadsheet shows very clearly that the GTX780 had +38mV added and the voltage and columns for the R9 290 is actually blank. They didn't touch voltage on the R9 290 and they used stock voltage increase on the GTX780.

Shocking !

 

[Fastx]

I owned 2x MSI TFII GTX 780 Gaming cards and in order to get over 1200 core I need to use the voltage modded BIOS. I also tested one with 1280 core at 1.28v but felt that as they were on air this was not sustainable.

IMHO people should expect ~1180 with stock BIOS and ~1220-1240 with the modded BIOS. Anything over this will be seen as a bonus.

But I am hoping for possibly one of the below EVGA GTX 780 SC ACX overclocks on a stock BIOS. but really whatever I get will be fine with me, but I can hope.

EVGA GTX 780 SC ACX 1254
http://www.guru3d.com/articles_pages/evga_geforce_gtx_780_sc_acx_review,26.html


EVGA GTX 780 SC ACX 1241
http://www.overclockers.com/evga-gtx-780-superclocked-acx-graphics-card-review

@jj109
Thanks

 

[Silverforce11]

Silicon Lottery time!!

Over 4000 samples: GTX 780 http://hwbot.org/hardware/videocard/geforce_gtx_780/

Over 200 samples: R290 http://hwbot.org/hardware/videocard/radeon_r9_290/

Feel free to add some extra to boost clocks in-game for the 780. Seems like a repeat of 7970 vs 680, peak (outside of freak silicon winners) for 7970 was around 1.25ghz, and for the 680, was around 1.33ghz.

 

[jj109]

I owned 2x MSI TFII GTX 780 Gaming cards and in order to get over 1200 core I need to use the voltage modded BIOS. I also tested one with 1280 core at 1.28v but felt that as they were on air this was not sustainable.

IMHO people should expect ~1180 with stock BIOS and ~1220-1240 with the modded BIOS. Anything over this will be seen as a bonus.

I think MSI's been dumpster diving to drive the costs down :biggrin:. All the TF 780 Gamings I've seen in reviews have trouble overclocking.

Silicon Lottery time!!

Over 4000 samples: GTX 780 http://hwbot.org/hardware/videocard/geforce_gtx_780/

Over 200 samples: R290 http://hwbot.org/hardware/videocard/radeon_r9_290/

Feel free to add some extra to boost clocks in-game for the 780. Seems like a repeat of 7970 vs 680, peak (outside of freak silicon winners) for 7970 was around 1.25ghz, and for the 680, was around 1.33ghz.

Pointless when people just crank up the voltage as high as it needs to be stable and do a benchmark run.

 

[Termie]

But I am hoping for possibly one of the below EVGA GTX 780 SC ACX overclocks on a stock BIOS. but really whatever I get will be fine with me, but I can hope.

EVGA GTX 780 SC ACX 1254
http://www.guru3d.com/articles_pages/evga_geforce_gtx_780_sc_acx_review,26.html


EVGA GTX 780 SC ACX 1241
http://www.overclockers.com/evga-gtx-780-superclocked-acx-graphics-card-review

@jj109
Thanks

I see you have your new 780. If you have Crysis 3, I'd suggest you use that to test the overclock. I can run all day at 1254/7000 in 3DMark (which OC.com used to test for stability, and by extension, the easy games Guru3D used to test stability). It crashes within 45 seconds every time I try Crysis 3. It's stable at 1215/6800 in that game. Also, note on VRAM overclocking - it can be stable at clocks that actually have a negative impact. 7000 is slower in many games than 6800 without crashing.

 

[blackened23]

Oh my! Did you read the link you thankfully provided? +25% power limit and +106% power limit does not = voltage increase.

I was incorrect on the voltage. I have no problem admitting a mistake on my part, I missed it. That said, this doesn't change the fact that the vast majority of reviews published for overclocking on the 290X/290 vs GK110 heavily favor the latter for maximum overclocks, overclock scaling, and clockspeeds above stock.

So far this has proven the case with water reference vs reference, aftermarket GK110 vs aftermarket hawaii, and reference air vs reference air. Nearly every aftermarket 290/290X review and under water review as published by tech websites show these cards are having difficulty in the range of 125-200MHz above stock. Whereas GK110 has been published basically everywhere to hit 400MHz+ above stock (~1300) given a custom PCB such as the classified, HOF, or GHz edition windforce cards. The GHz 780ti in particular is so fast beyond stock that it passes the GTX 690 - it scales that well. Even the reference cards are hitting 300MHz+ above stock (1200MHz+).

Don't misunderstand. Overclocking alone doesn't make the 290 a bad card. That isn't what i'm saying, far from it. I didn't consider the reference a great card for reasons I won't even get into. As long as the reference issues are fixed and priced right, it has potential to be fantastic. An aftermarket 290 would be a great card at MSRP. I see the Tri-X 290 review which looks to be a great card, as it fixes all of the issues I had with reference. Performs great compared to stock 780 and is quiet at load. Great card at MSRP. Unfortunately, 290s at MSRP don't exist in the states, although one can hope that situation changes sometime soon now that LTC is in the dumpster.

 

[blackened23]

I see you have your new 780. If you have Crysis 3, I'd suggest you use that to test the overclock. I can run all day at 1254/7000 in 3DMark (which OC.com used to test for stability, and by extension, the easy games Guru3D used to test stability). It crashes within 45 seconds every time I try Crysis 3. It's stable at 1215/6800 in that game. Also, note on VRAM overclocking - it can be stable at clocks that actually have a negative impact. 7000 is slower in many games than 6800 without crashing.

I agree with this so so much. I've had successful overclocked valley runs, with the same overclocks being crushed by crysis 3.

Crysis 3 really is the overclock killer, if you're on the edge of stability. Very good OC stability test.

 

[Termie]

Blackened23 - just to be fair, to compare overclocks, you have to go by stock operating frequency. Anandtech has published the stock operating frequency of the 780 Ti here: http://www.anandtech.com/show/7492/the-geforce-gtx-780-ti-review/16

It's 1020MHz. The 290X is 1000MHz. That makes it pretty easy to compare them, actually. And overall, the 780 Ti does seem to overclock slightly better.

Interestingly, almost all Nvidia 28nm high-end chips hit around 1200-1300, and all AMD 28nm high-end chips hit 1150-1250. So with both starting at around 1000MHz, the 780Ti had an advantage this round. It was the opposite with 7970 vs. 680, where the 680 operated at 1110MHz or so at stock and the 7970 was at 925MHz.

 

[blackened23]

Blackened23 - just to be fair, to compare overclocks, you have to go by stock operating frequency. Anandtech has published the stock operating frequency of the 780 Ti here: http://www.anandtech.com/show/7492/the-geforce-gtx-780-ti-review/16

It's 1020MHz.

I was referring to the GTX 780. This is what *I* said:

Whereas GK110 has been published basically everywhere to hit 400MHz+ above stock (~1300) given a custom PCB such as the classified, HOF, or GHz edition windforce cards.

I wasn't specific in mentioning the GK110, but I was referring to the 780 which has been tested everywhere in terms of OC'ing. This is why I mentioned the HOF - as far as I know, the HOF doesn't exist as a TI variant. Not yet anyway. I later mentioned the 780ti to demonstrate the scaling of the GK110. But that sentence was referring to the GTX 780. Anyway. It is 400Mhz above stock. The HOF 780 achieved speeds in excess of 1300mhz at HardOCP. The lightning GTX 780 achieved 1333MHz. 433MHz above stock. The stock boost of the reference GTX 780 is 900MHz.

GTX 780 specifications:

http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-780/specifications


GTX 780 GPU Engine Specs:
2304CUDA Cores
863Base Clock (MHz)
900Boost Clock (MHz)
160.5Texture Fill Rate (billion/sec)
GTX 780 Memory Specs:


An overclock north of 1300mhz, which has been achieved by quite a few GTX 780s as tested being 24/7 stable by review websites, is more than 400MHz over stock.

I think the 780ti could do the same (in terms of OC headroom over stock) given that it is the same silicon - but fully unlocked. But we just don't know yet. Aftermarket 780ti cards are just now hitting the streets, and most of them haven't been extensively tested yet.

Interestingly, almost all Nvidia 28nm high-end chips hit around 1200-1300, and all AMD 28nm high-end chips hit 1150-1250. So with both starting at around 1000MHz, the 780Ti had an advantage this round. It was the opposite with 7970 vs. 680, where the 680 operated at 1110MHz or so at stock and the 7970 was at 925MHz.

I also agree with this. I thought the 7970 was an excellent overclocker, and better than the initial GTX 680 GK104 cards in that respect. I'm just not seeing the same "beyond stock" headroom on the Hawaii chip. Maybe a future custom PCB can change things, but that's how things are right now from published data. Now i'm not saying this makes Hawaii bad in aftermarket variants. Clearly the chip performs well in the 1100-1200MHz range.

 

[Termie]

Sorry to pick a fight (especially right after you kindly supported my statement about Crysis 3) - I just don't think the published boost clock is that meaningful given the dynamic nature of GPU Boost 2.0. It's a bit hard to pin down a reference 780 boost clock, but I'd say it's around 1000MHz, not 900MHz.

That doesn't take away from the fact that both the 780 and the 780 Ti have more OC headroom on average than the 290x, and perhaps around the same headroom as the 290 (haven't seen as many OC reviews of that card).

 

[Fastx]

Termie my card should be here Saturday per post #105 above just updated my sig a little early. Crysis 3 and BF4 are my main games I play so will test my overclocks in C3.
Thanks for the tips!