R9 290 or GTX780 ??

[desprado]

Well let's look at some results of oced aftermarket 780's in comparison to Titan. First, I think we can agree based on multiple review sites that the Titan is generally about 9-10% faster than the reference 780.

TechPowerUP - 9.2% faster (1600p)
Computer Base - 8% faster (1600p)
Hardware Canucks - 9.8% faster
Hardware France - 7.7% faster (1440p)
Anandtech - 10% faster

Reviews of various aftermarket 780's

Guru3D : Oced 780 Classified (1293/1710) - Average of 21% faster across three games than reference 780 or 11% faster than Titan.
Anandtech: Oced 780 Superclocked ACX (1215/1700) - Average of 5.5% faster than Titan across 5 games
Hardware Canucks: Oced 780 SC ACX (1183/1722) - Average of 5.8% faster across two games
Tech Buyers Guru: Oced 780 SC ACX (1215/1700) - Average of 20% faster than reference 780 or 10% faster than Titan.
Guru3D : Oced 780 Windforce (1202/1693) - Average of 4% faster than Titan across 3 games.
Hardware Canucks: Oced 780 Windforce (1153/1714) - Average of 3.5% faster across 2 games.
Guru3D : Oced Asus 780 Poseidon (1241/1652) - Average of 12% faster than Titan across 3 benches.
PC Max: Oced 780 DC II (1205/1590) - Average of 7% faster than Titan across 3 benches
Xbit Labs: Oced 780 HerculeZ X3 (1150/1790) - They actually had results for 6 different overclocked 780s. The fastest out-of-the-box was the Hercules. It was 1% slower than the Titan on average. All 6 780's overclocked to 1124-1163Mhz which is only about 5% faster than stock in the case of the HerculeZ. Even with memory overclocks in the 10-15% range, it's very doubtful any of the cards are more than 5% faster than the Titan.
Hardware Overclock: Oced Palit 780 Super Jetstream (~1240/1624) - Average of 11% faster than Titan across 3 games.
Hardware Canucks: Oced 780 Lightning (1280/1612) - Average of 13.7% faster than Titan across 2 games.
Hardware Canucks: Oced 780 Classified (1320/1856) - Average of 15% faster than Titan across 2 games.
Guru3D : Oced 780 Lightning (1333/1710) - Average of 11% faster than Titan across 3 games.
Hardware Canucks: Oced 780 DC II (1213/1752) - Average of 8.1% faster than Titan across 2 games.
TechPowerUP: Six oced aftermarket 780s - The average overclock of the six aftermarket 780's was ~1200Mhz.

HardOCP: Oced 780 HoF (1300/1575) - Average of 21% faster than Titan across 4 games

[H]'s review is an outlier amongst all the other reviews above. An oced aftermarket 780 is around 8-9% faster than Titan on average.



From the relatively few 290X reviews around

Hardwareluxx: Oced 290X DC II (1195/1545) - Average of 22% faster than Titan across 3 games.
ComputerBase: 290X DC II and Gigabyte (1000-1050/1250-1350) - Both cards at stock clocks on average are 14.8% faster than the Titan across 5 games.
Hardware Canucks: Oced 290X DC II (1195/1504) - Average of 22.8% faster than Titan across 2 games.
Guru3D : Oced 290X DC II (1175/1546) - Average of 16% faster than Titan across 3 games.

I'd say the results are looking more favorable on the aftermarket 290X's than the 780's so far.

Op is asking about GTX 780 and R9 290 non X version

 

[Elfear]

Op is asking about GTX 780 and R9 290 non X version

Subtract 5-6% from the 290X data above then.

 

[desprado]

Subtract 5-6% from the 290X data above then.

Than GTX 780 OC is better than R9 290 OC
Enough said.

 

[Elfear]

Than GTX 780 OC is better than R9 290 OC
Enough said.

Uhhhh. Did you read my post? An oced 290X is ~20% faster than Titan on average so take 6% off and an oced aftermarket 290 is ~14% faster. An oced aftermarket 780 is ~8-9% faster than Titan so I'm not sure where you see oc 780 > oc 290.

 

[desprado]

Uhhhh. Did you read my post? An oced 290X is ~20% faster than Titan on average so take 6% off and an oced aftermarket 290 is ~14% faster. An oced aftermarket 780 is ~8-9% faster than Titan so I'm not sure where you see oc 780 > oc 290.

U did see the Topics where they compared GTX 780 OC and R9 290 OC and GTX clearly had about 4 to 5% advantage and u can check reviews if u want they also say the same.

 

[Elfear]

U did see the threads where they compared GTX 780 OC and R9 290 OC and GTX clearly had about 4 to 5% advantage and u can check reviews if u want they also say the same.

You mean like the 20-odd reviews I linked above?

 

[BallaTheFeared]

I pulled up the guru lightning review and the only of score they have is overall for firestike...

Should I look at more, kind of confused?

 

[Elfear]

I pulled up the guru lightning review and the only of score they have is overall for firestike...

Should I look at more, kind of confused?

I linked to the wrong page of the review but it's fixed now. There are oc results for Bioshock, Hitman, and Tomb Raider.

 

[Silverforce11]

Well let's look at some results of oced aftermarket 780's in comparison to Titan. First, I think we can agree based on multiple review sites that the Titan is generally about 9-10% faster than the reference 780.

Great post, with proper reviews to back it up.
So that settles it. Outside of fancy modded bios with vcore to the extreme, 780s seem to indeed average out at around 1.2ghz to 1.25ghz and is ~10% faster than Titan.

 

[BallaTheFeared]

1330 classified is slower than a 780 ti at stock? I don't buy it, but hey that just makes card that much more awesome.

Still as quick as your lightning at 1400 in bioshock and able to beat 1200mhz 290x in amd favored tests so either I'm doing something right or they're doing it wrong.

 

[Silverforce11]

1330 classified is slower than a 780 ti at stock? I don't buy it, but hey that just makes card that much more awesome.

Still as quick as your lighting at 1400 in bioshock and able to beat 1200mhz 290x in amd favored tests so either I'm doing something right or they're doing it wrong.

Yup, reviewers are doing it wrong and Balla is doing it right.

ps. When you say either, it tends to be one or the other, here, both scenarios = you being right, because if they are wrong, you're right... pretty arrogant, no? The more humble statement would be "either I'm doing something wrong or they are".

 

[BallaTheFeared]

Sounds about right.


According to this the 780 isn't even close to the ti at 1333/1900, also 780 scaling is purely random. Higher clock cards slower than lower clocked cards...

Oh and my card at 1300 has no trouble what so ever beating a stock 780 ti.

Also this:

 

[Silverforce11]

Sounds about right.

Okay, Balla and ABT is right and every other reviewers are wrong.. we got it.

 

[wand3r3r]

Well let's look at some results of oced aftermarket 780's in comparison to Titan. First, I think we can agree based on multiple review sites that the Titan is generally about 9-10% faster than the reference 780.

TechPowerUP - 9.2% faster (1600p)
Computer Base - 8% faster (1600p)
Hardware Canucks - 9.8% faster
Hardware France - 7.7% faster (1440p)
Anandtech - 10% faster

Reviews of various aftermarket 780's

Guru3D : Oced 780 Classified (1293/1710) - Average of 21% faster across three games than reference 780 or 11% faster than Titan.
Anandtech: Oced 780 Superclocked ACX (1215/1700) - Average of 5.5% faster than Titan across 5 games
Hardware Canucks: Oced 780 SC ACX (1183/1722) - Average of 5.8% faster across two games
Tech Buyers Guru: Oced 780 SC ACX (1215/1700) - Average of 20% faster than reference 780 or 10% faster than Titan.
Guru3D : Oced 780 Windforce (1202/1693) - Average of 4% faster than Titan across 3 games.
Hardware Canucks: Oced 780 Windforce (1153/1714) - Average of 3.5% faster across 2 games.
Guru3D : Oced Asus 780 Poseidon (1241/1652) - Average of 12% faster than Titan across 3 benches.
PC Max: Oced 780 DC II (1205/1590) - Average of 7% faster than Titan across 3 benches
Xbit Labs: Oced 780 HerculeZ X3 (1150/1790) - They actually had results for 6 different overclocked 780s. The fastest out-of-the-box was the Hercules. It was 1% slower than the Titan on average. All 6 780's overclocked to 1124-1163Mhz which is only about 5% faster than stock in the case of the HerculeZ. Even with memory overclocks in the 10-15% range, it's very doubtful any of the cards are more than 5% faster than the Titan.
Hardware Overclock: Oced Palit 780 Super Jetstream (~1240/1624) - Average of 11% faster than Titan across 3 games.
Hardware Canucks: Oced 780 Lightning (1280/1612) - Average of 13.7% faster than Titan across 2 games.
Hardware Canucks: Oced 780 Classified (1320/1856) - Average of 15% faster than Titan across 2 games.
Guru3D : Oced 780 Lightning (1333/1710) - Average of 11% faster than Titan across 3 games.
Hardware Canucks: Oced 780 DC II (1213/1752) - Average of 8.1% faster than Titan across 2 games.
TechPowerUP: Six oced aftermarket 780s - The average overclock of the six aftermarket 780's was ~1200Mhz.

HardOCP: Oced 780 HoF (1300/1575) - Average of 21% faster than Titan across 4 games

[H]'s review is an outlier amongst all the other reviews above. An oced aftermarket 780 is around 8-9% faster than Titan on average.



From the relatively few 290X reviews around

Hardwareluxx: Oced 290X DC II (1195/1545) - Average of 22% faster than Titan across 3 games.
ComputerBase: 290X DC II and Gigabyte (1000-1050/1250-1350) - Both cards at stock clocks on average are 14.8% faster than the Titan across 5 games.
Hardware Canucks: Oced 290X DC II (1195/1504) - Average of 22.8% faster than Titan across 2 games.
Guru3D : Oced 290X DC II (1175/1546) - Average of 16% faster than Titan across 3 games.

I'd say the results are looking more favorable on the aftermarket 290X's than the 780's so far.

Facts & data ftw.

Thanks for taking the time. It's annoying to see every thread turn into a contest and some appear to have taken the [h] HOF result and ran with it.

Personally I'm very interested in these comparisons (290 vs 780 & oc etc.) but it's like reading one sided marketing posts with all the best points of their cards being demonstrated by certain posters (intentionally or not).

Watch that goalpost shift in 3.2.1 now that yet another claim has been debunked...

We know the average 780 oc isn't the same as a few cherry picked lightning/classified/GHz modded chips. Yes it's a decent card, but not the 290-beater certain posters dream and post that it is.

Personally I still think it *might* be a slightly better OCer, basically only once using a warranty voiding mod but even then I'm not sure. It's mainly just a few posters repeating it over and over despite the custom 290's just starting to trickle out.

Yup, reviewers are doing it wrong and Balla is doing it right.

ps. When you say either, it tends to be one or the other, here, both scenarios = you being right, because if they are wrong, you're right... pretty arrogant, no? The more humble statement would be "either I'm doing something wrong or they are".

It's not even about humility, it's about rationality which appears to be of little concern for certain posters when trying to market their cards.

 

[desprado]

Okay, Balla and ABT is right and every other reviewers are wrong.. we got it.

Can u stop trolling plz.

 

[BallaTheFeared]

I know you want to ride this hard with you red flag waving. However I'll take logic and results that make sense over pretty much anything.

 

[Silverforce11]

I know you want to ride this hard with you red flag waving. However I'll take logic and results that make sense over pretty much anything.

There is no flag waving, I respect the 780 a lot for being an OC champ.. what I am disgusted with, is when you are faced with a massive wall of real cold hard facts backed up by very reputable hardware websites in their business for many years that is contrary to your own data, you have the arrogance to turn around and say they they are doing it wrong and you are right.

Disgusting.

 

[BallaTheFeared]

There is no flag waving, I respect the 780 a lot for being an OC champ.. what I am disgusted with, is when you are faced with a massive wall of real cold hard facts backed up by very reputable hardware websites in their business for many years that is contrary to your own data, you have the arrogance to turn around and say they they are doing it wrong and you are right.

Disgusting.

Tell me in so many words how this:

Beats this:

Also you can stop acting now, it wasn't a very good performance.

 

[Silverforce11]

Obviously you don't feel NV have improved their drivers since the 780 to the 780 ti launch to improve performance in games?

The only thing that didn't change, relative performance versus Titan.

 

[BallaTheFeared]

Obviously you don't feel NV have improved their drivers since the 780 to the 780 ti launch to improve performance in games?

The only thing that didn't change, relative performance versus Titan.

I thought of that of course, but it doesn't check out.

Any other way to explain why a card with the same uarch, with more bandwidth, more pixel/texture fill is slower than a slower card from the same uarch?

 

[BallaTheFeared]

Let me put it another way.

The 780 Ti needs to gain 7% more performance to match the 1200/1500 290x in Crysis 3.

How much more texture/pixel fill and bandwidth does the Classified at those clocks have over the stock 780 Ti?

 

[wand3r3r]

Let me put it another way.

As I've followed this the average 780 oc > 290 oc hasn't even really been proven.

The average 780 oc is not 20% over the titan as proven above.

The 780 Ti needs to gain 7% more performance to match the 1200/1500 290x in Crysis 3.

So it's close in Crysis 3. What is the point exactly?

I am not even sure if you are trying to demonstrate the 780 superiority over the 290 or the 780 ti atm. or if you're doing something else entirely?

It appears to be shifting to demonstrate the 780 supremacy in some other form now that it's oc myths being perpetuated are debunked (on average)?

 

[Leadbox]

Tell me in so many words how this:

Beats this:

576 more CC will stomp your OCed 780 (depending on game)

Let me put it another way.

The 780 Ti needs to gain 7% more performance to match the 1200/1500 290x in Crysis 3.

How much more texture/pixel fill and bandwidth does the Classified at those clocks have over the stock 780 Ti?

You're assuming that thats some sort of bottleneck on the stock Ti in that title?

 

[rtsurfer]

Tell me in so many words how this:

Beats this:

Also you can stop acting now, it wasn't a very good performance.

The Ti is faster.

But it has sone weird thing with its memory. I remember reading the 780 Ti owners thread on Overclock.net in its initial days of launch where owners of even classified 780 Ti were complaining about the memory being slower then their 780s.

But in gaming performance the Ti is still better.

 

[BallaTheFeared]

There seems to be some confusion with my results due to "skynet". That confusion seems to come from what it does, and why it's used.

What do the skynet bios do?

Well quite simply, they disable boost, increase power target, and unlock up to 1.21v.

I only use them for the power target unlock. The MSI AB softmod is what allows me to use 1.3v.

Well Balla, your 780 is only faster because you use modded bios and that is unrealistic and/or too enthusiast.

Fair point, pressing 1 to disable the bios protection and 3 to flash is complicated stuff. So in the interest of appeasing those with issues, I will bench my 780 with stock bios. I will however retain the MSI Afterburner softmod which allows up to 1.3v without any issues related to potentially voiding a warranty.

Barring something amazing happening, and we get some actual R290 series user in here there isn't a lot for us to go on.

We have one 290x at 1200/1375, represented by ICDP. I think everyone should thank him and our other 290 series user Gloomy who is submitting the 290 results at 1150/1500.

With only two games there isn't a lot to go on, hopefully we can come together as a community to do this right and get more games tested. However despite that limitation lets move forward with our user submitted results and do our comparisons.

Tomb Raider: 1920x1080 "Ultimate" preset

290 - 1150/1500 - 88.3 FPS

290x - 1200/1375 - 96.3 FPS

780 - 1280/1750 - 93.9 FPS*


Metro Last Light: 1920x1080 Very High/Normal/Very High

290 - 1150/1500 - 79.62 FPS

290x - 1200/1250 - 83.88 FPS

780 - 1293/1750 - 84.22 FPS*

*GTX 780 GHz Edition benched with Stock bios using MSI Afterburner soft mod to unlock voltage.


What can we take from these results?


Metro Last Light Stock vs Stock

In stock configuration the R290 has a 13% performance advantage over the 780 in Metro LL at these settings. This performance gap represents a deficit that isn't reflected in overall performance difference charts, as well as being the largest deficit shown on the Anand review of the Tri-X. If we assume the same performance jump for each card in other titles, like Crysis 3, or BF3 the 780 should pull ahead based on reason and logic alone.

Hopefully we can get some more testers, and more games done. I'd rather we do this ourselves. I think as a community we can and should come together to do this :thumbsup:

As I've followed this the average 780 oc > 290 oc hasn't even really been proven.

So long as only one R290x and one R290 show up to actually bench their computers it's going to remain that way.

I can't force R290 series users to put up results for comparison.

The average 780 oc is not 20% over the titan as proven above.

I think any critical thinker would call those results into question, considering there isn't even a consensus among the results themselves as far as where to place it. Not only that but the results look like lame duck overclocks, meaning the performance of the 780 in the tests don't line up with each other, let alone what users like myself get at similar clock speeds.

So it's close in Crysis 3. What is the point exactly?

Point was the performance of the Classified didn't match the performance of it's clocks. I thought that was fairly clear from my post?

I am not even sure if you are trying to demonstrate the 780 superiority over the 290 or the 780 ti atm. or if you're doing something else entirely?

Both really. First there is the issue of where to place the 780 OC, using two or three cards makes honing that down easier. Plus having the second Nvidia card in the mix allows us to see actual scaling of the core and uarch, which is why the results posted don't line up with actual performance expectations of the clocks reported. It shouldn't take much thought to figure out a 1300MHz classified is faster than a stock 780 Ti, despite what some reviews might have you believing.

It appears to be shifting to demonstrate the 780 supremacy in some other form now that it's oc myths being perpetuated are debunked (on average)?

No we're still on 780 OC > R290, we just got side tracked a bit by odd ball results which don't match expectations for the clocks reported.

576 more CC will stomp your OCed 780 (depending on game)

Wut?

You're assuming that thats some sort of bottleneck on the stock Ti in that title?

Of course.

This is like your troll comment on the Metro graph, you probably won't have much to say like after I posted the actual IG FRAPS frametimes.

The Ti is faster.

This is based on what exactly? The fact that the Ti has yet to dethrone age old 780 and Titan results on various websites?

But it has sone weird thing with its memiry. I remember reading the 780 Ti owners thread on Overclock.net in its initial days of launch where owners of even classified 780 Ti were complaining about the memory being slower then their 780s.

Ok?

But in gaming performance the Ti is still better.

Despite it losing to 780s and Titans in actual user benchmarks? Weird stuff, must just be a magic chip. Defying the laws of reason and logic, with less fill/bandwidth and still faster. Must be that magic vram everyone is complaining about. :thumbsup: