-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Galaxy GeForce GTX 650 Ti For Sale At Newegg
Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
no doubt the 7850 will be faster, but the most important factor will be the price,
also while bandwidth might be a big problem in some games, I'm not convinced that at 80+GB/s it will be as drastic as what happens to the GT 640, sure it's a much faster GPU which demands more bandwidth, but 80+GB/s is a massive improvement, and a good amount higher than the 7770 (the card which have lower memory bandwidth than a 5770, but is clearly faster, also the 7770 can beat the 256 bit 6790/5830 in most games)
overclocking varies, but I'm interested to see, I've seen some GTX 660 (with the same GK106) near 7GHz, if the memory and PCB on the 650 TI are good enough... there is certainly good potential for gain.
Red Hawk
Diamond Member
Way back when the Radeon HD 5770 was released, its 128 bit memory bus width was already bottlenecking it in games. 80+ GB/s is not a "massive improvement"; extra shader power just means that the 650 Ti is going to hit those bottlenecks even harder. I'm interested to see the benchmarks where the 7770 beats the 6790/5830 "in most games".
The GTX 660 has a memory clock of about 6000 MHz, while the 650 Ti is going to have a memory clock of 5400 MHz. As bandwidth is certainly going to be the limiting factor on this card, one would think that Nvidia would push the memory clock speed higher if there was consistent headroom for it. Judging by the size of the PCB I wouldn't put too much faith in it for OCability.
The GTX 660 has a memory clock of about 6000 MHz, while the 650 Ti is going to have a memory clock of 5400 MHz. As bandwidth is certainly going to be the limiting factor on this card, one would think that Nvidia would push the memory clock speed higher if there was consistent headroom for it. Judging by the size of the PCB I wouldn't put too much faith in it for OCability.
compare the 5770 to the 6790, there is hardly a huge difference in performance, for a much higher bandwidth,
GT 640 bandwidth is only 28GBps... that's a massive difference, sure the extra ALUs and TMUs and all of that will need more bandwidth, but it's a more acceptable level,
something with significant impact might also be 16 ROPs only (just like a 7770), but still... GTX 650 TI will have 64TMUs, the 7770 only have 40... I would agree that some of this advantage will probably be limited by other aspects (like bandwidth), but I would still expect a big advantage in many cases.
about 7770 (72GB/s) vs 6790 (134GB/s)
7770 wins in most games
http://ht4u.net/reviews/2012/msi_ge...LAA&filter[2][]=Post+AA&filter[3][]=16&aa=all
as for the 5770, the 7770 proves that the 5770 is a lot more limited by other aspects than by memory bus/bandwidth.
Red Hawk
Diamond Member
There is a difference in performance between the 5770 and the 6790. Not always incredible difference, because the 6790 had the same amount of shaders running at a slightly lower clock speed than the 5770, but there were differences across the board, sometimes by more than 25%:
Link because url is crazy long (the 6770 is the same thing as the 5770)
That just goes to demonstrate that there were significant bottlenecks thanks to the 76.8 GB/s bandwidth that upping the bandwidth alleviated. The 650 Ti is bound to run into the same problems. Will it be as bad as the 5770? Of course not. The 7770 is better than the 5770 (and 6790) at compute, tessellation, and other DirectX 11 operations. When the bottleneck is one of those, the 7770 excels, as will the 650 Ti. Note, though, that while the link you posted shows the 6790 managing to win a significant portion of the games over the 7770, bump the 6790 down to a 6770:
url again hidden because it's crazy long
and the 6770 wins even less, only managing to win over Dragon Age II (more of an anomaly, really), and loses by an even higher percentage in most of the other games. Again, indicating that memory bandwidth makes a significant difference.
It doesn't matter if the 650 Ti has a "massive difference" over the 640. The 640 is borderline not even useful for gaming at all; what matters are comparisons to the Radeon HD 7700 series and the standard GTX 650. The 650 Ti is going to hit the same memory bottlenecks as those. I mentioned the 640 because it illustrates the situation; the 640 has the potential to rival the 7750, but it doesn't even come close because of its limited memory speed. The difference in bandwidth between the 640 and the 7750 is similar to the difference between the 650 Ti and the 7850.
Last edited:
forget about Dragon Age 2 from this test, because there is clearly something going wrong (probably driver related);
I understand how you are using the 640 as an example, but I disagree that anything similar to that can happen, simply because at 28GB/s is just to little, but at 86 there is a lot more "room", the GTX 650 (GT 640 with DDR5) clearly have more memory bandwidth than needed (80GB/s), the 650 Ti will be somewhat limited, but not nearly as much as the 640 imho (it's basically 2x the GT 640 GPU but with 3x the memory bandwidth),
even though bandwidth is closer to the 7770 I expect the 650 Ti to be a closer to the 7850 than the 7770 in many cases.
about the 6790 vs 6770 you can't forget that the 6790 was significantly faster in some cases (like tessellation) since it's basically a 6800, when the 6770 is a 5770, but again the 7770 is clearly faster than the 5770 with lower memory bandwidth, I don't see why the 650 Ti can't be a good amount faster, it already have 15% higher memory bandwidth and in other aspects a lot more.
let's wait and see.
I understand how you are using the 640 as an example, but I disagree that anything similar to that can happen, simply because at 28GB/s is just to little, but at 86 there is a lot more "room", the GTX 650 (GT 640 with DDR5) clearly have more memory bandwidth than needed (80GB/s), the 650 Ti will be somewhat limited, but not nearly as much as the 640 imho (it's basically 2x the GT 640 GPU but with 3x the memory bandwidth),
even though bandwidth is closer to the 7770 I expect the 650 Ti to be a closer to the 7850 than the 7770 in many cases.
about the 6790 vs 6770 you can't forget that the 6790 was significantly faster in some cases (like tessellation) since it's basically a 6800, when the 6770 is a 5770, but again the 7770 is clearly faster than the 5770 with lower memory bandwidth, I don't see why the 650 Ti can't be a good amount faster, it already have 15% higher memory bandwidth and in other aspects a lot more.
let's wait and see.
BenSkywalker
Diamond Member
If the specs in the OP were right, for $150 it seems like it will slot closer in performance to a 7850 then the 7770 which considering the marketplace seems to be a very solid offering. I'm sure AMD will respond with another round of price cuts if that is the case, but it seems like it should be a strong part.
Side note- with the GPU specs seems like vendors could offer 'OC' versions of this part that will be major value leaders, pair it with some higher clocked RAM and it should be nipping at the 660's heels, they price it so it will scare off the 'casual' shopper and let the enthusiast bang4buck crowd get a good deal.
Red Hawk
Diamond Member
I disagree that the 650 Ti will be closer to the 7850 than the 7770. It'll be much like the situation with the 660 Ti, only worse. The 660 Ti has the same core count and clock speed as the 670. The 670 general beats or ties the Radeon HD 7950 at stock speeds. Chop out a portion of the memory bandwidth from the 670 to make the 660 Ti, and some of those advantages disappear. What's more, the 7870 is competitive with the 660 Ti in a fair amount of games, even beating it sometimes. With the 650 Ti, they're both cutting the core count down from the 660 GTX and lobotomizing the memory bandwidth to an even greater extent than between the 670 and 660 Ti. We've got a repeat of the situation with the 660 Ti or worse, calling it now.
As for the 6790 -- the 7770 has better tessellation than both the 5770 and 6790. Stepping up from the 5770 to a 6790 in comparison to the 7770 should not be enough change the winner if the deciding factor is tessellation. Memory bandwidth has to be the source of the issue. Don't get me wrong, the 650 Ti is going to be faster in games that aren't memory constrained, but in games that are it's going to hit a wall and hit it hard.
in many games which the 7770 beat the 6790 tessellation is also not a factor.
it's just a more efficient architecture in many ways and a faster GPU... the 6790 doesn't really need 133GB/s...
the 7770 can hit a wall for lack of power in other aspects to (40TMUs, 640GCN shaders), so there is always a compromise, but I think the 650 Ti is a lot better balanced than the GT 640 (less than twice the GPU with 3x the memory bandwidth)... and in many games the 7850 is probably far from really needing all the memory bandwidth at its disposal (the clearly faster 7870 have the same memory bandwidth...), so I would think in many or some cases the 650 Ti will com close, and at the worst cases it will still be clearly ahead of the 7770 (much stronger GPU, 20% more memory bandwidth)....
edit: I guess we will continue to disagree, the arguments are exposed, but it's better to wait for the final result in a few days (or weeks?)
Last edited:
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
