To get our weekly geekiness quota out of the way early, the desktop video card industry is a lot like The Force. There are two sides constantly at odds with each other for dominance of the galaxy/market, and balance between the two sides is considered one of the central tenants of the system. Furthermore when the system isn’t in balance something bad happens, whether it’s galactic domination or uncompetitive video card prices and designs.

To that end – and to bring things back to a technical discussion – while AMD and NVIDIA’s ultimate goals are to rule the video card market, in practice they serve to keep each other in check and keep the market as a whole balanced. This is accomplished by their doing what they can to offer similarly competitive video cards at most price points, particularly the sub-$300 market where the bulk of all video card sales take place. On the other hand when that balance is disrupted by the introduction of a new GPU and/or new video card, AMD and NVIDIA will try to roll out new products to restore that balance.

This brings us to the subject of today’s launch. Friday saw the launch of AMD’s Radeon HD 7790, a $149 entry-level 1080p card based on their new Bonaire GPU. AMD had for roughly the last half-year been operating with a significant price and performance gap between their 7770 and 7850 products, leaving the mid-$100 market open to NVIDIA’s GTX 650 Ti. With the 7790 AMD finally has a GTX 650 Ti competitor and more, and left unchallenged this would mean AMD would control the market between $150 and $200.

NVIDIA for their part has no interest in letting AMD take that piece of the market without a fight, and as such will be immediately countering with a new video card: the GTX 650 Ti Boost. Launching today, the GTX 650 Ti Boost is based on the same GK106 GPU as the GTX 650 Ti and GTX 660, and is essentially a filler card to bridge the gap between them. By adding GPU boost back into the mix and using a slightly more powerful core configuration, NVIDIA intends to plug their own performance gap and at the same time counter AMD’s 7850 and 7790 before the latter even reaches retail. It’s never quite that simple of course, but as we’ll see the GTX 650 Ti Boost does indeed bring some balance back to the Force.

NVIDIA GPU Specification Comparison
	GTX 660	GTX 650 Ti Boost	GTX 650 Ti	GTX 550 Ti
Stream Processors	960	768	768	192
Texture Units	80	64	64	32
ROPs	24	24	16	16
Core Clock	980MHz	980MHz	925MHz	900MHz
Boost Clock	1033MHz	1033MHz	N/A	N/A
Memory Clock	6.008GHz GDDR5	6.008GHz GDDR5	5.4GHz GDDR5	4.1GHz GDDR5
Memory Bus Width	192-bit	192-bit	128-bit	192-bit
VRAM	2GB	1GB/2GB	1GB/2GB	1GB
FP64	1/24 FP32	1/24 FP32	1/24 FP32	1/12 FP32
TDP	140W	134W	110W	116W
GPU	GK106	GK106	GK106	GF116
Architecture	Kepler	Kepler	Kepler	Fermi
Transistor Count	2.54B	2.54B	2.54B	1.17B
Manufacturing Process	TSMC 28nm	TSMC 28nm	TSMC 28nm	TSMC 40nm
Launch Price	$229	$149/$169	$149	$149

When NVIDIA produced the original GTX 650 Ti, they cut down their GK106 GPU by a fairly large degree to reach the performance and power levels we see with that card. From 5 SMXes and 3 ROP/Memory partitions, GK106 was cut down to 4 SMXes and 2 ROP partitions, along with having GPU boost removed and overall clockspeeds lowered. In practice this left a pretty big gap between the GTX 650 Ti and the GTX 660, one which AMD’s 7850 and now their 7790 serve to fill.

Despite the name GTX 650 Ti Boost, it’s probably more meaningful to call NVIDIA’s new card the GTX 660 light. The GTX 650 Ti Boost restores many of the cuts NVIDIA made for the GTX 650 Ti; this latest 650 has the core clockspeed, memory clockspeed, GPU boost functionality, and ROP partitions of the GTX 660. In fact the only thing differentiating the GTX 660 from the GTX 650 Ti Boost is a single SMX; the GTX 650 Ti Boost is still a 4 SMX part, and this is what makes it a 650 in NVIDIA’s product stack (note that this means GTX 650 Ti Boost parts will similarly have either 2 or 3 GPCs depending on which SMX is cut). Because clockspeeds are identical to the GTX 660, the GTX 650 Ti Boost will be shipping at 980MHz for the base clock, 1033MHz for the boost clock, and 6GHz for the memory clock.

The result of this configuration is that the GTX 650 Ti Boost is much more powerful than the name would let on, and in practice is closer to the GTX 660 in performance than it is the GTX 650 Ti. Compared to the GTX 650 Ti, the GTX 650 TI Boost has just 106% of the shading/texturing/geometry throughput, but due in large part to the return of the 3^rd ROP partition, ROP throughput has been boosted to 159%. Meanwhile thanks to the combination of higher memory clocks and the full 192bit memory bus, memory bandwidth has been increased to 166% of the GTX 650 Ti’s. Or compared to a GTX 660, the GTX 650 Ti Boost has 100% the ROP throughput, 100% the memory bandwidth, and 80% of the shading/texturing/geometry performance. The end result being that in memory/ROP bound scenarios performance will trend close to the GTX 660, while in shader/texture/geometry bound situations performance will easily exceed the GTX 650 Ti’s performance by 6-16%, depending on where GPU boost settles at.

Of course GTX 660-like performance does come with some tradeoffs. While the GTX 650 Ti was a 110W TDP part, the GTX 650 Ti Boost will be a 134W part, just shy of the 140W GTX 660. The GTX 650 Ti Boost runs at the same clockspeeds and the same voltages with the same amount of RAM as the GTX 660, meaning the power savings are limited to whatever power is saved from fusing off that SMX, which in practice will not be all that much. Even by NVIDIA's own reckoning they're minimal. So what we’re effectively looking at is a somewhat slower GTX 660 operating at near-GTX 660 power levels.

Driving home the point that the GTX 650 Ti Boost is a reconfigured GTX 660, with the TDP being held at 140W NVIDIA and their partners will be recycling their GTX 660 designs for NVIDIA’s new card. Our reference card is identical to our GTX 660 reference card, and the same can be said for many partner designs. Partners need to provide the same power and cooling to the GTX 650 Ti Boost as they do the GTX 660, so there’s little point in rolling new designs and in fact this helps NVIDIA and their partners get the GTX 650 Ti Boost to market sooner.

Moving on to the launch and pricing details, as with NVIDIA’s other GK106 card launches, this is a pure virtual launch with partners rolling their custom designs from day one. Partners will be shipping both stock clocked and overclocked cards, and the mix should be very similar to what we saw with the GTX 660’s launch. Unexpectedly, partners will also have the option of going with 1GB or 2GB cards. 1GB cards are a late addition to NVIDIA’s lineup, in what appears to be an attempt to get down to price parity with AMD’s 7790. Our reference card is a 2GB model, and like other mixed cards like AMD’s 7850 we expect most GTX 650 Ti Boost cards to be 2GB cards.

In something of a coup for NVIDIA, 2GB GTX 650 Ti Boost cards will be hitting stores this week at $169, nearly a week ahead of the 7790. So despite launching second in this latest scuffle, NVIDIA will be the first to hit the market. Meanwhile 1GB cards will arrive later, hitting the market in early April at $149. Overclocked cards will of course carry their own premiums.

The competition for the GTX 650 Ti Boost will include a number of cards from both NVIDIA and AMD. The tight pricing of the market between $149 and $199 means that the GTX 650 Ti Boost will in practice be pulling double-duty as a 7790 and a 7850 competitor. Priced at $149, the later to arrive 1GB cards are the true 7790 competitor in every sense of the word.  Meanwhile with the 2GB cards launching at $169 they’re more akin to a 7850 competitor, something NVIDIA doesn’t hesitate to point this out.

In any case, the GTX 650 Ti Boost will be boxed in by the GTX 650 Ti below it at around $130, meanwhile above it will be the 7850 2GB at around $185 and the GTX 660 at around $200. Ultimately NVIDIA has to be sure to cover $149 to go up directly against the 7790, but if you can afford to spend a little more than $149, then between factory overclocked cards and new models there’s a different card at every $10.

Finally, like AMD, NVIDIA will be extending their promotional bundling to their latest card. The GTX 650 Ti Boost will quality for the same $75 in free-to-play game credits as the GTX 650 Ti, split up as $25 for World of Tanks, $25 for Hawken, and $25 for Planetside 2.

Spring 2013 GPU Pricing Comparison
AMD	Price	NVIDIA
	$209	GeForce GTX 660
Radeon HD 7850 2GB	$184
	$169	GeForce GTX 650 Ti Boost 2GB
Radeon HD 7790	$149	GeForce GTX 650 Ti Boost 1GB
	$134	GeForce GTX 650 Ti
Radeon HD 7770	$109	GeForce GTX 650
Radeon HD 7750	$99	GeForce GT 640

 

The Test

Since the reference GTX 650 Ti Boost is completely identical in build to the reference GTX 660, let’s jump right into our tests.

NVIDIA’s launch drivers for the GTX 650 Ti Boost are the recently released 314.21 driver set, which as an aside we’ve discovered fix Titan’s OpenCL issues and we’ll be following up on that next month. Meanwhile for the rest of our cards we’ll be reusing our data from last week’s 7790 launch, so 314.21 for the other NVIDIA cards, AMD driver 12.101.2 for the 7790, and Catalyst 13.2 B7 for the rest of the AMD cards.

Please note that the GTX 650 Ti Boost NVIDIA is sampling is the 2GB card. We’ll take a look at 1GB cards later once those arrive, since they were a late addition.

On a side note, by the time you’re reading this, this will be the third major GPU article we’ve posted in as many business days, and we still have one more to go tomorrow. So we apologize in advance for not having had the time to run any additional NVIDIA cards besides the 600 series; we’ll get that caught up in Bench once this batch of launches and conferences is over.

CPU:	Intel Core i7-3960X @ 4.3GHz
Motherboard:	EVGA X79 SLI
Power Supply:	Antec True Power Quattro 1200
Hard Disk:	Samsung 470 (256GB)
Memory:	G.Skill Ripjaws DDR3-1867 4 x 4GB (8-10-9-26)
Case:	Thermaltake Spedo Advance
Monitor:	Samsung 305T
Video Cards:	AMD Radeon HD 7850 (2GB) AMD Radeon HD 7790 AMD Radeon HD 7770 AMD Radeon HD 6870 NVIDIA GeForce GTX 650 Ti Boost (2GB) NVIDIA GeForce GTX 660 NVIDIA GeForce GTX 650 Ti
Video Drivers:	NVIDIA ForceWare 314.21 AMD 12.101.2 7790 Press Beta AMD Catalyst 13.2 Beta 7
OS:	Windows 8 Pro

 

DiRT: Showdown

Racing to the front of our benchmark suite is our racing benchmark, DiRT: Showdown. DiRT: Showdown is based on the latest iteration of Codemasters’ EGO engine, which has continually evolved over the years to add more advanced rendering features. It was one of the first games to implement tessellation, and also one of the first games to implement a DirectCompute based forward-rendering compatible lighting system. At the same time as Codemasters is by far the most prevalent PC racing developers, it’s also a good proxy for some of the other racing games on the market like F1 and GRID.

DiRT’s advanced lighting engine isn’t really within reach of all of these sub-$200 cards, but once we drop down to standard lighting there’s performance to spare, even with MSAA. Though even without that advanced lighting though this is one of the harder games for the GTX 650 Ti Boost; it falls behind the 7790 by about 5%, never mind the 7850.

Total War: Shogun 2

Our next benchmark is Shogun 2, which is a continuing favorite to our benchmark suite. Total War: Shogun 2 is the latest installment of the long-running Total War series of turn based strategy games, and alongside Civilization V is notable for just how many units it can put on a screen at once. Even 2 years after its release it’s still a very punishing game at its highest settings due to the amount of shading and memory those units require.

Shogun at Ultra quality has been favoring our NVIDIA cards as of late. As a result the GTX 650 Ti Boost becomes the cheapest card to cross the 30fps threshold at these Ultra settings. Though at the more meaningful very high quality settings, we get something that’s more an archetypical outcome for these cards, with the GTX 650 Ti Boost in close pursuit of the 7850 and clearly ahead of the 7790. This is a shader bound game, so compared to the vanilla GTX 650 Ti, the boost variant picks up some performance, but it’s still not enough to even halve the gap to the GTX 660.

Hitman: Absolution

The third game in our lineup is Hitman: Absolution. The latest game in Square Enix’s stealth-action series, Hitman: Absolution is a DirectX 11 based title that though a bit heavy on the CPU, can give most GPUs a run for their money. Furthermore it has a built-in benchmark, which gives it a level of standardization that fewer and fewer benchmarks possess.

Hitman is another game that sees the GTX 650 Ti Boost struggle compared to AMD’s cards, and also another game to have NVIDIA’s relative performance drop as the game’s quality settings are reduced. The GTX 650 Ti Boost is fast enough to hit 50fps at High settings, still trailing the 7850 by 10%. But if we want to get above 60fps we have to drop to Medium, at which point performance is closer to the 7790 than it is the 7850.

Meanwhile compared to the GTX 650 Ti and GTX 660, this is about as normal of a game as we’ll see. The GTX 650 Ti Boost greatly benefits from the additional ROP/memory resources compared to the GTX 650 Ti, pushing performance ahead by 30%, but lacking that 5^th SMX it will still trail the GTX 660 by 14% or so. Most games are shader/texture/geometry bound to the point that the GTX 660, while close to the GTX 650 Ti Boost, will never really be threatened.

Looking at our minimum framerates, the relative rankings are a bit more in NVIDIA’s favor. The GTX 650 Ti Boost still trails in the same spots, but not by as much as was the case with the average framerates.

Sleeping Dogs

Another Square Enix game, Sleeping Dogs is one of the few open world games to be released with any kind of benchmark, giving us a unique opportunity to benchmark an open world game. Like most console ports, Sleeping Dogs’ base assets are not extremely demanding, but it makes up for it with its interesting anti-aliasing implementation, a mix of FXAA and SSAA that at its highest settings does an impeccable job of removing jaggies. However by effectively rendering the game world multiple times over, it can also require a very powerful video card to drive these high AA modes.

With Sleeping Dogs we once again have to drop to normal AA to get above 60fps, which is a simple FXAA mode with no SSAA. The GTX 650 Ti Boost can’t catch the 7850 here, but it can get close, within 5%. Meanwhile we know from history that this game has a lot going on shader-wise, so it’s no great surprise that the gains relative to the GTX 650 Ti aren’t quite as great here as they are elsewhere, with the GTX 650 Ti Boost improving by 25%.

Minimum framerates are not in NVIDIA’s favor here. While the GTX 650 Ti Boost can approach the 7850 on average, at the most demanding points in the game performance drops to something a lot closer to the 7790.

Crysis: Warhead

Up next is our legacy title for 2013, Crysis: Warhead. The stand-alone expansion to 2007’s Crysis, at over 4 years old Crysis: Warhead can still beat most systems down. Crysis was intended to be future-looking as far as performance and visual quality goes, and it has clearly achieved that. We’ve only finally reached the point where single-GPU cards have come out that can hit 60fps at 1920 with 4xAA.

Crysis has shown to favor raw ROP performance and memory bandwidth above shader performance, so it’s been a rough game for NVIDIA’s Kepler cards, which typically have less memory bandwidth than their AMD competition. In this case at Gamer quality the GTX 650 Ti Boost still can’t crack 60fps, and it’s trailing the 7850 by 12%. On the other hand it has 50% more memory bandwidth than the 7790, so the 1GB variant may be interesting to see here if it can achieve similar results as the 2GB variant.

It’s interesting to note though that because this game depends so much on ROP performance and memory bandwidth that this is as close as we’ll see the GTX 650 Ti Boost get to the GTX 660. There’s less than 5% separating them at Gamer quality; these cards have plenty of shading/texturing performance, but not enough ROP performance to satisfy Crysis.

Jumping to our minimum framerates, we do see things open up a bit as shader bottlenecking does occur in a few places. The result is that the GTX 650 Ti Boost falls well behind the 7850 in minimums, and the full GK106 GTX 660 pulls ahead. Still, the difference between the GTX 650 Ti and its boost variant is nothing short of staggering; the boost card leads by 45% here, relying heavily on those ROP and memory bandwidth advantages.

Far Cry 3

The next game in our benchmark suite is Far Cry 3, Ubisoft’s island-jungle action game. A lot like our other jungle game Crysis, Far Cry 3 can be quite tough on GPUs, especially with MSAA and improved alpha-to-coverage checking thrown into the mix. On the other hand it’s still a bit of a pig on the CPU side, and seemingly inexplicably we’ve found that it doesn’t play well with HyperThreading on our testbed, making this the only game we’ve ever had to disable HT for to maximize our framerates.

At medium-high quality, the GTX 650 Ti Boost is just fast enough to crack 60fps on Far Cry 3.This puts it in a cozy position just below the 7850, trailing by under 5%, and well ahead of the 7790. This is a shader-heavy game though, so the gains over the GTX 650 Ti are limited to a bit over 20%, while the GTX 660 is ahead by almost another 20%.

Battlefield 3

Our final action game of our benchmark suite is Battlefield 3, DICE’s 2011 multiplayer military shooter. Its ability to pose a significant challenge to GPUs has been dulled some by time and drivers, but it’s still a challenge if you want to hit the highest settings at the highest resolutions at the highest anti-aliasing levels. Furthermore while we can crack 60fps in single player mode, our rule of thumb here is that multiplayer framerates will dip to half our single player framerates, so hitting high framerates here may not be high enough.

Our rule of thumb here is that we have to hit 60fps on this benchmark to ensure we don’t bottom out below 30fps in multiplayer, so surprisingly Ultra quality with just FXAA is quite playable here. Kepler cards simply excel at Battlefield 3, making this the one game that the GTX 650 Ti Boost can beat the 7850 at, and quite handily too at nearly 15%. Even dropping the quality to Medium for better performance still doesn’t dent the GTX 650 Ti Boost’s lead here. Meanwhile BF3 is another game that makes good use of shading and texturing, so the Boost’s gains over the GTX 650 Ti are around 27% or so, while the GTX 660 is still ahead by a further 13%.

Civilization V

Our final game, Civilization V, gives us an interesting look at things that other RTSes cannot match, with a much weaker focus on shading in the game world and a much greater focus on creating the geometry needed to bring such a world to life. In doing so it uses a slew of DirectX 11 technologies, including tessellation for said geometry, driver command lists for reducing CPU overhead, and compute shaders for on-the-fly texture decompression.

Civilization V is an interesting game due to the fact that it puts the 7790, GTX 650 Ti Boost, and 7850 so close together in performance. It stresses just about everything at some point – tessellation/geometry, ROP throughput, compute, and texturing – but it’s really shading/texturing that form the biggest bottleneck here. This works out well enough for NVIIDA, allowing them to get within 4% of the 7850, while still keeping the GTX 650 Ti Boost and GTX 660 well separated.

Compute Performance

As always we'll start with our DirectCompute game example, Civilization V, which uses DirectCompute to decompress textures on the fly. Civ V includes a sub-benchmark that exclusively tests the speed of their texture decompression algorithm by repeatedly decompressing the textures required for one of the game’s leader scenes.  While DirectCompute is used in many games, this is one of the only games with a benchmark that can isolate the use of DirectCompute and its resulting performance.

As our Civilization V compute benchmark is just that, a compute benchmark, so our results aren’t too surprising. This is one of the few compute tests NVIDIA does well at, so the GTX 650 Ti Boost is close to both Radeon cards, and not all that far behind the GTX 660 either.

Our next benchmark is LuxMark2.0, the official benchmark of SmallLuxGPU 2.0. SmallLuxGPU is an OpenCL accelerated ray tracer that is part of the larger LuxRender suite. Ray tracing has become a stronghold for GPUs in recent years as ray tracing maps well to GPU pipelines, allowing artists to render scenes much more quickly than with CPUs alone.

Moving on to LuxMark, we quite frankly transition into a more normal compute benchmark pattern for NVIDIA, which sees Kepler flopping. The GTX 650 Ti Boost can’t get even remotely close to a 7770, let alone the 7850. On the NVIDIA side it doesn’t help that since this is a compute benchmark the GTX 650 Ti Boost gains fairly little over the GTX 650 Ti.

Our 3rd benchmark set comes from CLBenchmark 1.1. CLBenchmark contains a number of subtests; we’re focusing on the most practical of them, the computer vision test and the fluid simulation test. The former being a useful proxy for computer imaging tasks where systems are required to parse images and identify features (e.g. humans), while fluid simulations are common in professional graphics work and games alike.

CLBenchmark is much the same as LuxMark, with NVIDIA cards bringing up the rear. The fluid simulation ends up being the more painful of the two benchmarks for the GTX 650 Ti Boost, clocking in at less than 1/3^rd the performance of the 7850.

Moving on, our 4th compute benchmark is FAHBench, the official Folding @ Home benchmark. Folding @ Home is the popular Stanford-backed research and distributed computing initiative that has work distributed to millions of volunteer computers over the internet, each of which is responsible for a tiny slice of a protein folding simulation. FAHBench can test both single precision and double precision floating point performance, with single precision being the most useful metric for most consumer cards due to their low double precision performance. Each precision has two modes, explicit and implicit, the difference being whether water atoms are included in the simulation, which adds quite a bit of work and overhead. This is another OpenCL test, as Folding @ Home is moving exclusively OpenCL this year with FAHCore 17.

NVIDIA still struggles at compute with FAHBench – the move to OpenCL isn’t doing them any favors – but it’s not the blowout that was our last two benchmarks. Interestingly explicit favors NVIDIA more than implicit, which may mean NVIDIA is handling the overhead better than AMD is. Still, any Folding @ Home users will be far better served by AMD than NVIIDA here.

Our 5^th compute benchmark is Sony Vegas Pro 12, an OpenGL and OpenCL video editing and authoring package. Vegas can use GPUs in a few different ways, the primary uses being to accelerate the video effects and compositing process itself, and in the video encoding step. With video encoding being increasingly offloaded to dedicated DSPs these days we’re focusing on the editing and compositing process, rendering to a low CPU overhead format (XDCAM EX). This specific test comes from Sony, and measures how long it takes to render a video.

Vegas is another OpenCL benchmark, and another benchmark NVIDIA brings up the rear with. Certainly the additional compute performance of the GTX 650 Ti Boost over the GTX 650 Ti is helping NVIDIA here, but it can’t make up for a gap of over 30 seconds.

Wrapping things up, our final compute benchmark is an in-house project developed by our very own Dr. Ian Cutress. SystemCompute is our first C++ AMP benchmark, utilizing Microsoft’s simple C++ extensions to allow the easy use of GPU computing in C++ programs. SystemCompute in turn is a collection of benchmarks for several different fundamental compute algorithms, as described in this previous article, with the final score represented in points. DirectCompute is the compute backend for C++ AMP on Windows, so this forms our other DirectCompute test.

SystemCompute mixes things up a bit with its multiple sub-benchmarks, but it still doesn’t change the fact that Kepler and GTX 650 Ti Boost just don’t do that well in most compute scenarios. 68K points is enough to tie the 6870 of all things, itself not a particular good compute card. Otherwise the bar is set by AMD at over 100K points.

Synthetics

As always we’ll also take a quick look at synthetic performance to get a better look at our video cards' underpinnings. These tests are mostly for comparing cards from within a manufacturer, as opposed to directly comparing AMD and NVIDIA cards.

We’ll start with 3DMark Vantage’s Pixel Fill test, a mix of a ROP test and a bandwidth test to see if you have enough bandwidth to feed those ROPs.

3DMark Vantage’s pixel fill test confirms what we know from the specs of the GTX 650 Ti Boost: that it has received a massive boost in ROP performance and memory bandwidth. The 45% greater pixel throughput rate here doesn’t reach the kind of lofty goals that the theoreticals would put it at, but it’s clearly quite an improvement. Interestingly despite the equal ROP throughput and memory bandwidth of the GTX 660 and GTX 650 Ti Boost, the GTX 660 is still clearly in the lead here. We’ve never looked at the impact of GPCs here, so if our card is a 2 GPC model then this might explain what we’re seeing.

Moving on, we have our 3DMark Vantage texture fillrate test, which does for texels and texture mapping units what the previous test does for ROPs.

Texture fillrates on the other hand are really only benefitting from the higher clockspeeds of the GTX 650 Ti Boost over the GTX 650 Ti, and memory bandwidth to a much lesser extent. This is why despite the similarities between the GTX 650 Ti Boost and the GTX 660, the latter is still quite safe from the GTX 650 Ti Boost.

Finally we’ll take a quick look at tessellation performance with TessMark. We have everything turned up to maximum here, which means we're looking at roughly 11 million polygons per frame.

NVIDIA has always had a fairly ridiculous geometry throughput rate, and that doesn’t change on the GTX 650 Ti Boost. A score of 753 is second only to the GTX 660, and well ahead of the 7850, which is an interesting confluence of a 2 primative/clock rate, and its lower clockspeeds relative to the 7790 and GTX 650 Ti Boost.

Power, Temperature, & Noise

Last but not least of course is our look at power, temperature, and noise. Because the GTX 650 Ti Boost is essentially a reconfigured GTX 660 there aren’t going to be any grand revelations here, but it will be useful to figure out just what the real-world power savings will be from fusing off that one SMX.

GeForce GTX 650 Ti Boost Voltages
Ref GTX 650 Ti Boost Idle	Ref GTX 650 Ti Boost Load	Ref GTX 660 Load
0.887v	1.175v	1.175v

With a peak load voltage of 1.175v, the GTX 650 Ti Boost tops out at the same voltage as the rest of the boost-enabled Kepler family.

Up next, before we jump into our graphs let’s take a look at the average core clockspeed during our benchmarks. Because GTX 650 Ti Boost has the same TDP has the GTX 660 but at least marginally lower power consumption due to the disabled SMX, it’s in an interesting position where it has more headroom for boosting than its fully-enabled counterpart. As a result we’re seeing far less variability than what we saw with the GTX 660 when we reviewed it last year. With the exception of BF3, every game is sustained at the top boost bin of 1071MHz. Based on these results it would appear that in practice the GTX 650 Ti Boost operates at a marginally higher average clockspeed than the otherwise superior GTX 660.

GeForce GTX 650 Ti Boost Average Clockspeeds
Max Boost Clock	1071MHz
DiRT:S	1071MHz
Shogun 2	1071MHz
Hitman	1071MHz
Sleeping Dogs	1071MHz
Crysis	1071MHz
Far Cry 3	1071MHz
Battlefield 3	1058MHz
Civilization V	1071MHz
FurMark	992MHz

Starting as always with idle power, 110W at idle is par for the course for most NVIDIA cards. With the reactivation of the 3^rd ROP partition the GTX 650 Ti Boost gives up the slight advantage the GTX 650 Ti gained here.

Moving on to our first and arguably most important load power test we have BF3. Power consumption in BF3 scales slightly with performance due to the extra work required of the CPU to feed more frames to a video card, but it usually paints some clear trends and this is no exception. NVIDIA may only be giving the GTX 650 Ti Boost a official TDP 6W lower than the GTX 660, but it’s clear power consumption is at least a little bit lower than that; we’re seeing 289W at the wall versus 298W for the GTX 660. On the other hand these results are 28W higher than the GTX 650 Ti, and even 18W higher than the 7850, never mind the 7790. The GTX 650 Ti Boost’s performance here is well ahead of everything other than the GTX 660, so this jump in power consumption does come with a performance boost, but it serves as a reminder that there is a tradeoff to be made. In the case of the GTX 650 Ti Boost, we’re seeing it pull away a bit from the efficiency curve set by NVIDIA’s other products.

As for power consumption under FurMark, we’re seeing the more muted results for the GTX 650 Ti Boost. Here it’s roughly halfway between the GTX 650 Ti and GTX 660, and probably should be a bit higher still. The fact that there’s a difference (at the wall) of 27W between the GTX 660 and GTX 650 Ti Boost is more than we would except, and more than makes sense for cards that are identical except for a single fused off SMX. So we may also be seeing chip-to-chip variation play a part here. In any case, power consumption is also similar to the 7850, but this is one of those scenarios where we put more faith in the BF3 numbers than the FurMark numbers; NVIDIA appears to just outright be more aggressive on throttling here.

Utilizing the same cooler as the GTX 660, there’s no surprise here in seeing the GTX 650 Ti Boost hit the same 30C idle temperatures.

Similarly, our temperature results here closely parallel the GTX 660’s under load. The GTX 650 Ti Boost consumes a bit less power than the GTX 660, and doesn’t get quite as warm as a result. The large jump from the GTX 650 Ti comes as a bit of a shock to the eyes at first, but as the GTX 650 Ti Boost is a blower and one with a conservative fan curve at that, this is to be expected.

With FurMark we see temperatures go up, but for the most part things are consistent with what we saw under BF3. The larger gap between the GTX 650 Ti Boost and GTX 660 reflects the larger gap in power consumption we saw earlier.

Just as with idle temperatures, the same GPU on the same cooler means we’re looking at the same idle noise temperatures. The blower NVIDIA uses here is quite good, but it can’t compete with simple open-air coolers like the 7790’s.

There’s a clear cutoff here between the open-air coolers and the blowers. The GTX 650 Ti Boost is quite a bit faster than something like the 7790 here, but the difference in noise is remarkable. Some of NVIDIA’s partner cards should fare much better here as they pack open-air coolers, with the usual tradeoff of giving up being fully-exhausting coolers. Still, this is a reminder that the GTX 650 Ti Boost pulls back from the efficiency curve a bit; it’s about 15% slower than the GTX 660 but no quieter for it.

Finally we have load noise under FurMark. NVIDIA’s more aggressive throttling here means that our results don’t jump up to much from BF3. The GTX 650 Ti Boost does finally end up being a bit quieter than the GTX 660 due to the former’s lower power consumption, and this is the only time we see the GTX 650 Ti Boost gain an edge on the 7850 in power/temp/noise.

Final Words

Bringing this review to a close, given the back-to-back launches of the GeForce GTX 650 Ti Boost and the Radeon HD 7790, our first instinct is to frame the GTX 650 Ti Boost in reference to the 7790. Indeed the GTX 650 Ti Boost will be NVIDIA’s 7790 competitor, but what we’re reviewed – a 2GB GTX 650 Ti Boost – is not the same as the 1GB card that will occupy that $149 price point. So we want to hold off on that comparison for a bit. We need to see a 1GB GTX 650 Ti Boost to properly make that evaluation.

When we opened this article we mentioned how there’s practically a card at every $10 between $100 and $200. For consumers on a fixed budget this is great since it means there’s a video card at every price, but without distinct islands it makes it harder for us as reviewers to make a recommendation. But as always we’ll take a stab at it.

In our 7790 review we mentioned how uneasy we are with 1GB cards right now due to the fact that larger cards have been on the market for some number of years by now, and game consoles are about to take a massive leap in RAM capacities. Of course neither AMD nor NVIDIA is going to offer a balanced 2GB card at $149 right now – what you can have is a 1GB 7790 or a 1GB GTX 650 Ti Boost. So with that said we’re going to hold to our earlier recommendation that if you’re in the market for a card around these price ranges and you can afford to go past $149, we’re at a point where a 2GB card is a reasonable investment. And to that end the 2GB GTX 650 Ti Boost is going to be $10-$15 cheaper than the 2GB 7850, making it the more affordable option for a 2GB card.

Moving along, with the chief competitor for the 1GB GTX 650 Ti Boost being the recently launched Radeon HD 7790, the chief competitor for the 2GB version is going to be AMD’s Radeon HD 7850 2GB. The GeForce card will have a $10-$15 price advantage on average, keeping in mind that the price of the 7850 is typically in flux due to price changes and temporary rebates.

The performance of the two cards is close at times, but ultimately the GTX 650 Ti Boost spends most of its time trailing the 7850 to some degree, its only lead being in Battlefield 3. With those BF3 results pushing the cards to within 5% of each other on average, otherwise the gap is closer to 8%. As such the GTX 650 Ti Boost is unquestionably the weaker card in our benchmarks, just not significantly so. So for it to be priced under the 7850 is the right move here.

In any case, as a 7850 competitor the GTX 650 Ti Boost is nothing amazing – its price and performance are close to the 7850, a card that has been sitting at its current price for months now – meaning it fills its intended role as a slightly cheaper, slightly slower 7850 competitor, but nothing more. If we had to pick between the two of them the 7850 does look a bit better due to its slightly higher performance and lower power consumption, but most buyers should be happy with either one. As we said in our introduction this is a case of balance being returned to the market – where there was once just the 7850 there is now a viable alternative from NVIDIA in the form of the GTX 650 Ti Boost 2GB.

Ultimately with a card at every $10 amidst a packed market there isn’t any kind of real sweet spot right now, so we don’t have any strong recommendations. For buyers on a fixed budget recent launches like the GTX 650 Ti Boost and 7790 introduce newer and better options at $169 and $149 respectively. Otherwise there’s a clear chain of progression right up to $200, and the best card is going to be the card you can afford.