Is 2600k graphics better than X1900 XT?

[RoyG]

I used to play games with X1900 XT years ago. I have not played games for quite some time. I just built a PC with 2600k. Wonder if 2600k GPU performance is comparable to X1900 XT. Any idea?

 

[Red Hawk]

I don't think so. A X1900 XT is roughly comparable with an Xbox 360, which Sandy Bridge graphics can't quite match.

 

[GaiaHunter]

Most likely between GF 6600 GT and GF 7600GT.

The X1900 XT should be around ATi HD 4650

 

[Roy2001]

Most likely between GF 6600 GT and GF 7600GT.

Thanks, that's better than I expected. In windows 7 the score for graphics/3D is 5.8 which is very good IMO.

 

[Termie]

Most likely between GF 6600 GT and GF 7600GT.

The X1900 XT should be around ATi HD 4650

This. The x1900xt (which I once owned), is about equivalent to an HD4650. The 2600k's GPU is about equivalent to an HD5450, which is about half as fast as an HD4650. I agree that HD5450-class performance would be on par with a 7600gt (or x800xt), so that's a good old-school comparison to the 2600k.

Benchmarks of 2600k vs. HD5450: http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/4

 

[Red Hawk]

An X1900 XT is actually closer to a an HD 4670 due to memory bandwidth (4650 has 16 GB/s, 4670 has 32 GB/s, and the X1900 XT has 46.4 GB/s), and an HD 5450 is less than half as fast as a HD 4670 -- 80 shader processors and 12.8 GB/s bandwidth as opposed to the 4670's 320 shader processors and 32 GB/s bandwidth. So if we roughly equate a 2600k GPU to a HD 5450 and a X1900 XT to a HD 4670, the 2600K GPU comes out quite weak compared to the X1900 XT.

 

[Nemesis 1]

I used to play games with X1900 XT years ago. I have not played games for quite some time. I just built a PC with 2600k. Wonder if 2600k GPU performance is comparable to X1900 XT. Any idea?

Also keep in mind you have the 2600k and that runs up to 1350 mgz so your scores would be better, These things also do pretty good @ o/c ANand results pretty much suck comparred to what were getting with the 2600k. I didn't see what memory was used in these setup nor have got the new drivers as of yet . We running low latencymemory at 1600 .

 

[Termie]

An X1900 XT is actually closer to a an HD 4670 due to memory bandwidth (4650 has 16 GB/s, 4670 has 32 GB/s, and the X1900 XT has 46.4 GB/s), and an HD 5450 is less than half as fast as a HD 4670 -- 80 shader processors and 12.8 GB/s bandwidth as opposed to the 4670's 320 shader processors and 32 GB/s bandwidth. So if we roughly equate a 2600k GPU to a HD 5450 and a X1900 XT to a HD 4670, the 2600K GPU comes out quite weak compared to the X1900 XT.

Interesting analysis based on theoretical numbers, but as it turns out, you're not quite correct. I also owned an HD4670 - it's significantly faster than an x1900xt.

Here are some benchmarks to make that clear:

(1) HD4670 versus HD3850: http://www.anandtech.com/show/2616/6
(2) HD3850 versus x1950xtx: http://www.anandtech.com/show/2376/9.

The 4670 is on par with a 3850, while the 3850 is about 50% faster than the x1950xtx, which itself was about 8-10% faster than the x1900xt: http://www.anandtech.com/show/2069/6

Thus, a 4670 is not a good comparison to the x1900xt - it's about 60% faster.

Aren't old reviews fun?

 

[GaiaHunter]

Interesting analysis based on theoretical numbers, but as it turns out, you're not quite correct. I also owned an HD4670 - it's significantly faster than an x1900xt.

Here are some benchmarks to make that clear:

(1) HD4670 versus HD3850: http://www.anandtech.com/show/2616/6
(2) HD3850 versus x1950xtx: http://www.anandtech.com/show/2376/9.

The 4670 is on par with a 3850, while the 3850 is about 50% faster than the x1950xtx, which itself was about 8-10% faster than the x1900xt: http://www.anandtech.com/show/2069/6

Thus, a 4670 is not a good comparison to the x1900xt - it's about 60% faster.

Aren't old reviews fun?

It is quite relevant these days, when people speculate about future APU products performance, to show that is quite possible to increase performance without having to increase bandwidth.

 

[Red Hawk]

Interesting analysis based on theoretical numbers, but as it turns out, you're not quite correct. I also owned an HD4670 - it's significantly faster than an x1900xt.

Here are some benchmarks to make that clear:

(1) HD4670 versus HD3850: http://www.anandtech.com/show/2616/6
(2) HD3850 versus x1950xtx: http://www.anandtech.com/show/2376/9.

The 4670 is on par with a 3850, while the 3850 is about 50% faster than the x1950xtx, which itself was about 8-10% faster than the x1900xt: http://www.anandtech.com/show/2069/6

Thus, a 4670 is not a good comparison to the x1900xt - it's about 60% faster.

Aren't old reviews fun?

If the increased memory bandwidth doesn't help the X1900 XT against the 4670 then the 4650 would be better as well. It's the same chip as the 4670, just with slower core clock speed and half the memory bandwidth.

 

[CFP]

I wonder how the Llanos compare. I used to own an X1900XTX. Yup, I paid that much more for 25MHz on the core.

Loved it.

 

[Zap]

Maybe an A8 Llano would be closer in GPU performance?

 

[MangoX]

Hmm I really wonder this as well, as I own a X1900GT, not a XT but close. I wonder how my card compares to the more recent low end offerings.

 

[Termie]

If the increased memory bandwidth doesn't help the X1900 XT against the 4670 then the 4650 would be better as well. It's the same chip as the 4670, just with slower core clock speed and half the memory bandwidth.

The 4650 is actually slightly faster, but not significantly. Check Tom's hierarchy charts: The 4650 DDR3 (which was rare) is ranked as equivalent to the x1950xtx: http://www.tomshardware.com/reviews/best-graphics-card-game-performance-radeon-hd-6670,2935-7.html. The DDR2 version would not be as fast. So in this case, memory bandwidth matters. It all comes down to balance - the x2900xt has loads of memory bandwidth but no balance...the HD3850 was much better balanced, and could come close to x2900xt performance with half the memory bandwidth, the same core, and slower core clocks:

"Historically, AMD's RV series has been a cost cut version of their R series designed for lower end volume parts, and that's where RV670 started. Right off the bat, half the external and internal memory bandwidth of R600 [2900xt] was cut out. External bandwidth dropped from 512-bit to 256-bit, but AMD stuck with 8 memory channels (each dropped from 64bit to 32bit).

Internally, the ring bus dropped from 1024-bit to 512-bit. This cut in bandwidth contributed to a significant drop in transistor count from R600's ~720M. RV670 is made up of 666M transistors, and this includes the addition of UVD hardware, some power saving features, the necessary additions for DX 10.1 and the normal performance tuning we would expect from another iteration of the architecture."

Source: http://www.anandtech.com/show/2376/1

Here's how performance stood: http://www.anandtech.com/show/2376/7

Again, never judge a GPU based on theoretical bandwidth alone - you'll come to the wrong conclusions half the time.

I wonder how the Llanos compare. I used to own an X1900XTX. Yup, I paid that much more for 25MHz on the core.

Loved it.

Maybe an A8 Llano would be closer in GPU performance?

The review I linked before was actually the Llano review. Here's another page of performance analysis: http://www.anandtech.com/show/4448/amd-llano-desktop-performance-preview/3. The Llano A8 gives performance equivalent to the HD6450 (despite having the core of an HD5570), and is thus about twice as fast as both the HD5450 and the 2600k. As Anandtech notes in this case, the A8 GPU is being held back by memory bandwidth - note how when the system memory is overclocked to DDR3-1866, the performance jumps 20-30%. So that just goes to show that memory bandwidth can be critical (the other half of the time...see above!).

Conclusion: Llano is almost as fast an an x1900xt. Note that in the Tom's link above, the HD6450 is ranked as equivalent to the x1950gt. This is where Llano would fall.