by Alexey Stepin , Yaroslav Lyssenko
11/30/2006 | 08:29 PM
The Nvidia GeForce 7950 GT graphics card was announced along with the GeForce 7900 GS, but was targeted at a higher price category, their recommended prices being $299 and $199, respectively. In the ATI camp, the release of the Radeon X1950 XTX brought about considerable changes into the high-performance product line-up. The mentioned card, priced at $449, became the new flagship with a new type of memory on board, replacing the Radeon X1900 XTX.
The Radeon X1900 XT, a lower-class device, suffered a graphics memory reduction from 512MB to 256MB, but acquired a much lower price, $259, instead. Having the same recommended price, the GeForce 7900 GT thus faced a formidable opponent that had much better technical specs.
Nvidia conceived its GeForce 7950 GT as a solution to win the $259-299 sector back. It didn’t take much effort to create that graphics card as Nvidia had already had the simple and cheap G71 processor and a PCB to match. The GPU and memory frequencies of the GeForce 7950 GT were easily set at 550MHz and 1400MHz, respectively, whereas the amount of graphics memory was set at 512MB.
As a matter of fact, the GeForce 7950 GT is identical to the GeForce 7800 GTX 512 in its specs, save for the memory clock rate (1400MHz against 1700MHz). You may remember that the 7800 GTX 512 used to be a top-end solution that once aspired after absolute leadership in the premium product class but was very rare due to limited production volumes.
The G71 being a very energy-efficient chip, the increase in the frequency didn’t call for any changes in the graphics card design. Thus, it won’t be a big mistake to say that the new card, just like the GeForce 7900 GS, is yet another variation of the GeForce 7900 GT model.
The manufacturer’s recommended price for the GeForce 7950 GT is set at $299, which is quite affordable for many PC gamers. A curious fact, a performance comparable to that of the 7950 GT could only be had for as much as $649 just a year ago!
The Radeon X1900 XT 256MB looks a dangerous rival to the GeForce 7950 GT, though. Its graphics core not only has a higher clock rate, but also surpasses the core of Nvidia’s card in computing power. Practice shows that the 48 pixel processors can indeed be helpful in today’s games whereas the small number of TMUs the R580 chip incorporates may be well compensated by the higher frequency they are clocked at. So, ATI’s strategy of putting the focus on high speed of execution of math1ematics-heavy pixel shaders is bringing its fruit now. This is the result of game developers’ making wider use of complex special effects like parallel occlusion mapping. We dwelt upon that topic at length in our reviews of the ATI Radeon X1900 XTX and Radeon X1950 XTX.
ATI’s solution may seem to be limited by having only 256 megabytes of memory, but there are not so many games today that need large amounts of it. You may only need that much if you play in high resolutions at extremely high levels of full-screen antialiasing, but in this case $299 products often lack muscle to deliver a comfortable speed (you will see if this statement is true in the Tests section of this review).
So, the $299 GeForce 7950 GT has an advantage in terms of memory amount but is weaker than its market opponent as concerns pixel shader performance.
Just as it was the case with the GeForce 7900 GS, the PCB design of the GeForce 7950 GT is an exact copy of the GeForce 7900 GT that we described in our earlier review.
Although the cards are made using the same reference PCB and carry identical coolers, Nvidia’s partners have more freedom with the GeForce 7950 GT than they had with the GeForce 7900 GTX and 7950 GX2, and original models have appeared on the market. For example, XFX’ version of the GeForce 7950 GT has a black PCB and a passive cooler. There also exist a number of pre-overclocked versions of the card.
The main frequency of the GeForce 7950 GT GPU that all its units, save for the vertex processors, are clocked at is 550MHz. The frequency delta is the same as with other Nvidia’s cards on the G71 chip, namely 20MHz. It means that the vertex processors are clocked at 570MHz. This is a rather big step forward over the GeForce 7900 GT, and the use of the old cooler may seem indiscreet. We will check this out in our power consumption tests.
The GeForce 7950 GT carries eight 512Mb GDDR3 chips (Infineon HYB18H512321AF-14) for a total of 512 megabytes of graphics memory. The chips work at 2.0V voltage and have a rated frequency of 700 (1400) MHz. This is indeed the frequency the chips are clocked at by the card, providing a bandwidth of 44.8GB/s across a 256-bit bus. The memory already working at its rated frequency, there’s little hope it will do well at overclocking, although there may be exceptions (for example, we once managed to achieve a 20% memory frequency growth from our GeForce 7900 GS).
The card has a standard connector configuration: two DIV-I ports capable of working in dual-link mode and a universal 7-pin S-Video/YPbPr port. Every GeForce 7950 GT supports HDCP (the card features a special chip that stores keys to decode protected HD video content).
The reference cooler of the GeForce 7950 GT consists of a copper base and a folded copper sheet glued to it. It is equipped with a 45mm 0.19A fan that has a two-wire connection to the PCB. The cooler, covered with a plastic casing, is fastened to the card with four spring-loaded screws and takes heat off the graphics core only; the memory chips get no cooling at all. Dark grey thermal grease is used as a thermal interface between the cooler and the GPU die.
There could not be any apprehensions about that rather simple cooling solution at GPU frequencies about 450MHz due to the low heat dissipation of the G71 chip, but it may prove inadequate for a frequency of 550MHz. Some graphics card manufacturers, particularly EVGA and XFX, seem to share this opinion as they equip their versions of the card with more advanced coolers.
We measured the level of noise produced by the graphics card’s cooler with a digital sound-level meter Velleman DVM1326 (0.1dB resolution) using A-curve weighing. At the time of our tests the level of ambient noise in our lab was 36dBA and the level of noise at a distance of 1 meter from a working testbed with a passively cooled graphics card inside was 40dBA. We got the following results:
Like the GeForce 7900 GS, the GeForce 7950 GT graphics card features a fan speed management system and thus produces little noise at work. The fan is working at its full speed only for the first couple of seconds after your turning the PC on. The system has two operation modes, a quiet 2D and a noisier 3D, but the GeForce 7950 GT remains virtually silent in every mode. The Nvidia card is preferable in this respect to the Radeon X1900 XT that comes with an old-type cooler that has been a cause of complaints from us as well as from users who want to have a quiet PC. The new cooler developed by ATI Technologies, free from the shortcomings of the older model, is only installed on Radeon X1950 XTX.
The power consumption of the Nvidia GeForce 7950 GT was measured according to our traditional methods on the following testbed:
The mainboard in this testbed is modified: measuring shunts are connected inline into the power lines of the PCI Express x16 slot. The shunts have connectors for measuring instruments. The 2xMolex → PCI Express power adapter is also equipped with such a shunt. We perform our measurements with a Velleman DVM850BL multimeter that has a measurement accuracy of 0.5%.
To create a 3D load on the graphics card, we ran the first SM3.0/HDR test from Futuremark 3DMark06 in a loop at 1600x1200 resolution with 16x anisotropic filtering. A Peak 2D load was created by launching the 2D Transparent Windows test from Futuremark PCMark05. Here are the results:
Click to enlarge
As might have been expected, the GeForce 7950 GT takes a place between the GeForce 7900 GT and GeForce 7900 GTX. Its power draw of 61.1W in 3D mode is quite a good result for a full-featured G71-based graphics card with 512 megabytes of graphics memory on board. The Radeon X1950 Pro consumes considerably more power.
The GeForce 7950 GT resembles the GeForce 7900 GT when it comes to loading the different power lines. That’s natural as they use the same PCB design. The +3.3V line bears almost no load at all while the +12V lines have the same loads in 3D mode. As opposed to that, the GeForce 7900 GTX puts a higher load on the internal +12V line that goes through the PCI Express x16 slot.
The GeForce 7950 GT has good power consumption properties, yet its reference cooler may turn to be ineffective on it in some situations. That’s why we think the decision of some of Nvidia’s partners to equip such graphics cards with more effective coolers, justifiable.
It had been reasonable to expect good overclocking from GeForce 7900 GT or GS cards because the GPU’s frequency potential had been far from exhausted, but we are far from optimistic about the GeForce 7950 GT model. 650MHz is obviously the practical limit for the G71 and the percentage of chips capable of working at such a high frequency is rather low. As opposed to that, almost each sample of the G71 is stable at 550MHz, but further overclocking depends entirely on your luck. We were rather lucky as we managed to increase the GPU clock rate to 610MHz and our card was stable at that, but we don’t expect each sample of the GeForce 7950 GT to be as good at overclocking as ours.
The memory chips showed much better overclockability, speeding up from 700 (1400) MHz to 790 (1580) MHz. So, although we didn’t surpass the memory frequency of the original GeForce 7800 GTX 512, we got very near to the level of the GeForce 7900 GTX.
Our overclocked GeForce 7950 GT turned to be much alike to the GeForce 7900 GTX, so we decided not to benchmark it in this mode. Its performance just wouldn’t differ much from that of the GeForce 7900 GTX.
We want to remind you that the stock cooler installed on the GeForce 7950 GT has too low performance to attempt serious overclocking. You should replace it or use additional cooling means to avoid damaging your graphics card at overclocked frequencies.
We testedthe performance of our Nvidia GeForce 7950 GT graphics card on the following platforms:
The drivers were set up in such a way as to provide the highest possible texture filtering quality.
We selected the highest possible graphics quality level in each game. We didn’t edit the games’ configuration files. The speed was measured using the game’s integrated tools or, if not available, by means of the Fraps program. We also measured the minimum speed of the cards where possible.
Besides the two standard resolutions of 1280x1024 and 1600x1200 pixels, we also used a resolution of 1920x1200 pixels with an aspect ratio of 16:10 in games supporting widescreen modes. We enabled FSAA and anisotropic filtering from the game’s menu. If this was not possible, we forced them using the appropriate driver settings.
We decided not to run any tests with disabled FSAA, because the technical specification of the graphics cards priced between $259 and $299 allow us to hope for relatively high performance even with FSAA 4x enabled. We ran the tests with disabled FSAA only for those games that do not support FSAA due to the specifics of their engine or use HDR (FP16). The thing is that the GeForce 7 family cannot perform FSAA along with floating-point HDR.
Unfortunately, at the time we were working on this article we didn’t have a Radeon X1950 XT 256MB at our disposal therefore we had to use a slightly more expensive model with 512MB of memory. Here is the full list of graphics cards, besides Nvidia GeForce 7950 GT that participated in our test session:
The following games and applications were used in this test session:
First-person 3D shooters:
Third-person 3D shooters:
The results of the GeForce 7950 GT are high enough for comfortable play in 1600x1200 and are higher than those of the Radeon X1950 Pro, but lower than the results of the top-end cards Radeon X1900 XT and GeForce 7900 GTX.
Note that we’ve included a Radeon X1900 XT with 512MB of graphics memory into this test session. The results of the 256MB version of the card may be lower, especially in high resolutions with enabled full-screen antialiasing.
The GeForce 7950 GT delivers the same performance as the Radeon X1900 XT in all the resolutions, which is a good result, but the average frame rate is too low to enable 4x FSAA in this game. Well, you can’t get a really comfortable speed in 1280x1024 even on the GeForce 7900 GTX here.
The Radeon X1950 Pro and GeForce 7950 GT suit equally well for running Far Cry at any resolution, including the yet-exotic 1920x1200. Nvidia’s card is somewhat faster in 1280x1024, though.
The same goes for the results the cards show on the Research map. The GeForce 7950 GT provides a considerable performance increase over the GeForce 7900 GT, about 25-30%.
But you should keep it in mind that the GeForce 7950 GT is more expensive than the Radeon X1950 Pro and does not provide as high anisotropic filtering quality as the latter does.
It’s all generally the same in the FP16 HDR mode as in the previous test, but you cannot use resolutions above 1280x1024 on a GeForce 7950 GT or on a cheaper Radeon X1950 Pro here. The Radeon X1900 XT and the GeForce 7900 GTX are the only graphics cards to ensure a comfortable average frame rate in the resolution of 1600x1200 pixels.
The speed of the Radeon X1950 Pro is probably limited by its TMUs and ROPs. The GeForce 7950 GT has more of such subunits and also a comparable core frequency and thus enjoys a 15-18% advantage over its rival. It makes the resolution of 1280x1024 playable with enabled FSAA.
F.E.A.R. is a very demanding application. It doesn’t run sufficiently fast even on the single Radeon X1900 XT in 1600x1200, although the latter leaves the GeForce 7950 GT quite far behind.
The deferred rendering technique employed in this game is incompatible with full-screen antialiasing. That’s why there are only anisotropic filtering mode results here.
The GeForce 7950 GT is a mere 4fps slower than the GeForce 7900 GTX. It makes an excellent choice for this game among single $300 cards because it outperforms the Radeon X1900 XT with 512MB of memory.
The Radeon X1950 Pro seems to suffer from poorly optimized drivers or from lack of TMUs and ROPs, yet it competes successfully with its market rival GeForce 7900 GS.
The Radeon X1900 XT and the GeForce 7950 GT both deliver good performance in Half-Life 2: Episode One , but ATI’s solution is somewhat faster and boasts a higher quality of anisotropic filtering. The Radeon X1950 Pro is limited to 1280x1024 if you enable 4x full-screen antialiasing.
You just can’t use 4x FSAA on a GeForce 7950 GT because this card doesn’t notch 50fps even in 1280x1024. The Radeon X1900 XT, on the contrary, is close to the necessary level of performance, although it fails in the higher resolutions, too.
Being less demanding than Prey , this game provides a more favorable environment for the graphics cards to race in. The GeForce 7950 GT shows an acceptable speed in 1600x1200, outperforming the Radeon X1950 Pro by 17-18%.
The Radeon X1900 XT feels at its ease even in 1920x1200, but do not forget that we are benchmarking its version with 512MB of graphics memory. The cheaper 256MB version may have produced different results.
Serious Sam 2 needs 512 megabytes of graphics memory to run fast. That’s why the GeForce 7950 GT is far ahead of the Radeon X1950 Pro in two out of the three resolutions, but is slower than the Radeon X1900 XT.
If even the X1900 XT cannot yield 60fps in 1280x1024, you should better give up using FSAA here. Your lowering the level of detail in the game settings would worsen the game’s visuals more.
The GeForce 7950 GT is close to delivering a speed of 40-45fps that is considered optimal for third-person shooters in 1280x1024. The card is competing with the Radeon X1900 XT and even shows better min fps numbers.
The results of the Radeon X1950 Pro are much more modest. You cannot run this game smoothly on it in 1280x1024, at least not with enabled FSAA and anisotropic filtering.
The Radeon X1950 Pro makes up for the lower performance of its TMUs and ROPs with its numerous pixel shader processors and even beats the GeForce 7950 GT as concerns minimum frame rates. The average frame rates of the two cards are similar. The Radeon X1900 XT is beyond competition here.
The GeForce 7950 GT is at first equal to the GeForce 7900 GTX but loses its ground due to slower memory as the resolution grows. Although this card provides an acceptable average frame rate in 1280x1024, its performance may decline dramatically, to 4-5fps, in complex scenes, spoiling your gaming experience.
However, this is the typical behavior of all the tested graphics cards: the game speed may bottom out to 10fps even on the Radeon X1900 XT.
There is no such great need for a fast graphics memory subsystem in closed scenes of this game as there is in open environments, so the GeForce 7950 GT is but slightly worse than the leading Radeon X1900 XT and GeForce 7900 GTX.
In open scenes, however, there is a bigger gap between the GeForce 7950 GT and 7900 GTX, at least in resolutions below 1920x1200. The Radeon X1950 Pro has got closer to the GeForce 7950 GT here and has even delivered higher min performance in some cases despite having only half the amount of graphics memory.
Among graphics cards from the $259-299 price range it is only the Radeon X1900 XT that is capable of delivering a comfortable frame rate in 1280x1024.
The problem we noticed with the GeForce 7900 GT and GS in our earlier tests is not observed with the GeForce 7950 GT and GeForce 7900 GTX, probably due to the larger amount of graphics memory. However, the results of these cards differ but little, so there is some limiting factor, perhaps in the ForceWare driver, that has nothing to do with GPU and memory frequencies.
Radeon X1000 graphics cards do not support vertex texturing and cannot use the game’s SM3.0 mode to render the water surface with the highest possible quality. This is only possible on Nvidia’s GeForce 7 architecture the game is optimized for.
Thanks to its more efficient OpenGL driver and to the specifics of the game engine, the GeForce 7950 GT easily beats the Radeon X1900 XT, providing a comfortable average speed in all the resolutions.
The GeForce 7950 GT is lagging behind the Radeon X1950 Pro in all the resolutions due to the preferences of the game engine, but both cards are fast enough to allow using FSAA at 1280x1024. The Radeon X1900 XT delivers a similar speed in 1920x1200, becoming an unrivalled leader in this test.
The GeForce 7950 GT is 8-10% behind the GeForce 7900 GTX regardless of the resolution, finishing third behind the Radeon X1900 XT. We’d recommend you to avoid using resolutions above 1280x1024 with enabled FSAA because the speed of the GeForce 7950 GT may bottom out to below 25fps in that case.
Although the GeForce 7950 GT is a little ahead of the Radeon X1950 Pro in the standard resolutions, they deliver identical performance in 1920x1200, probably because the ring-bus memory controller shows its very best in that case. It is also clear that the amount of graphics memory doesn’t affect the game speed much whereas the clock rates are vitally important.
Generally speaking, none of the graphics cards has a high speed here, so you’ll probably have to choose between FSAA and high resolutions.
Having more TMUs and ROPs and a comparable graphics core frequency, the Nvidia GeForce 7950 GT still cannot beat the cheaper Radeon X1950 Pro in 3DMark03, scoring 318 points less. There’s no talking about its challenging the Radeon X1900 XT here.
The GeForce 7950 GT enjoys the biggest advantage over the Radeon X1950 Pro in the first game test. This is expectable considering the specifics of the test scene and the importance of having a high fill rate to render it quickly. This advantage is smaller in the third test because ATI’s card can fully utilize its 36 pixel processors. In the second test the competing solutions from ATI and Nvidia match each other’s performance. As for the Radeon X1900 XT, it is unrivalled (the GeForce 7900 GTX outperforms it only once – in the first test’s 1920x1200 resolution).
The GeForce 7950 GT looks better than the Radeon X1950 Pro in the 4x FSAA + 16x AF mode, but it also costs considerably more money.
The GeForce 7950 GT scores more points in 3DMark06 than the Radeon X1950 Pro but 439 points less than the Radeon X1900 XT.
The difference is smaller in the SM2.0 tests that are free from advanced shaders, but bigger in the SM3.0/HDR tests where the progressive graphics architecture from ATI works under favorable conditions. The Radeon X1900 XT is just a little faster than the GeForce 7950 GT in the first group, but easily overtakes the GeForce 7900 GTX in the second group of tests.
The results of the separate SM2.0 tests look logical to us. The first test is a burden for the relatively weak TMU and ROP subsystems of the Radeon X1950 Pro. Well, even the Radeon X1900 XT, with much better specs, is a mere 0.2fps ahead of the GeForce 7950 GT.
The second test provides an opportunity for ATI’s cards to show their muscle on processing the complex geometry when rendering the abundant vegetation and to use the advanced lighting and shadows processing capabilities of the RV570 and R580 chips.
The GeForce 7950 GT looks generally preferable to the Radeon X1950 Pro but ATI’s solution belongs to a lower price category whereas the Radeon X1900 XTX, at least its version with 512MB of graphics memory, leaves no chance to the GeForce 7950 GT.
Boasting higher performance in comparison with the GeForce 7900 GT, the Nvidia GeForce 7950 GT graphics card hasn’t achieved a victory over its market opponents. So if you want to buy a GeForce 7950 GT or a same-class alternative from ATI, you should decide what games you are planning to run and check out the results of the cards in those games.
It’s hard to say anything about the GeForce 7950 GT in comparison with the Radeon X1900 XT because we don’t have a 256MB version of the latter card whereas its original version is equal to or better than the GeForce 7900 GTX. The Radeon X1900 XT 256MB is probably going to have the same performance except for extreme resolutions and FSAA and for games that need over 256MB of memory at the max level of detail. The Radeon X1900 XT 256MB also provides a high-quality anisotropic filtering that can only be matched by Nvidia’s expensive GeForce 8800 GTX and 8800 GTS card as yet. Moreover, the Radeon X1900 XT 256MB comes at a lower recommended price: $259 against the GeForce 7950 GT’s $299.
Anyway, GeForce 7950 GT is a good graphics card. Offering rather high performance, it is small, almost silent and power efficient which makes it more suitable than the Radeon X1900 XT 256MB for compact system cases or barebone systems. Its superb support for OpenGL must also be mentioned as it is important for OpenGL games like Prey, Quake 4 or Pacific Fighters.
The position of the GeForce 7950 GT on the market today seems somewhat insecure, but its price may decline dramatically after the arrival of the new Nvidia GeForce 8800 family. In this case the GeForce 7950 GX2 and 7900 GTX will get cheaper and push the GeForce 7950 GT into the market niche currently occupied by the GeForce 7900 GS. The Radeon X1950 Pro will have a dangerous rival then.