lethalkeyblade1

Hello non-existent people who may or may not (most likely the latter) care about my stuff
I will be returning with a redone version of my previous fic, Skull Heart
Because I'm bored and y'know, fixing janky dumb shiz into dumb shiz is a way to alleviate boredom
To everyone else, hope you're doing alright. 

Please note I am planning a video version of this review as well.

Hello all. Today I will be giving my review of the Nvidia GTX 960. A bit farfetched to do it now considering that it has been around for more than a year, and with AMD Polaris and Nvidia Pascal up on the horizon, how does this card hold up today? While I’ll be focusing on my KFA2 card, I’ll also do a general summary of the 960 cards as well.

Now, the model I have is the KFA2 reference cooler model. Aesthetically, simple, what you’d expect from a reference design card of course, usual 9.5 Inches length and dual slot, but the interesting thing is that the PCB is actually very small, and that the cooler is what’s big about it. The reference blower design also isn’t the most desirable, but you’d be surprised to know that this card handles temperatures very easily, and that it doesn’t whine very heavily unless set to high fan speeds above 70%. At least that’s my interpretation of it, but otherwise, a non-reference card would be a better choice.

My GTX 960, like I’ve mentioned, is a reference model from KFA2 with 4GB of GDDR5 VRAM.

Now the basic specs of the 960 of course are…

 

Shading Units: 1024
TMUs: 64
ROPs: 32
SMM Count: 8
GPU Clock: 1127Mhz
Boost Clock: 1178 Mhz
Memory Clock: 1753 Mhz, 7012Mhz Effective
Memory Size: 2048/4096 MB (2GB/4GB)
Memory Type: GDDR5
Memory Bus: 128 bit
Bandwidth: 112 GB/s

Now, it’s not exactly meant to wow like a 970 would, but this is what it is, a mid-range card specifically targeting mid-range performance.
With my model, strangely enough, default clocks are 1157Mhz, though not a game changing difference. I’ve previously done a review the GTX 750 Ti, and that card would be considered two steps back, with the 950 filling the gap. So, what does two steps above the 750 Ti mean? Well, let’s get into the benchmarks!

Test System is a Dell Precision T5500, so not exactly a typical built system today but it uses older hardware, so more or less performance should be more akin to a typical midrange PC today.
CPU: Xeon X5690 Hexa-Core (12 Threads)
RAM: 16GB DDR3 ECC 1333Mhz
SSD/HDD: Samsung 256GB SSD P50 Pro, 1TB Seagate Barracuda, 1TB Western Digital Black Enterprise
GPU: GTX 960 KFA2 4GB
PSU: Dell 825W Silver 
Operating System: Windows 10 Pro 

GPU is overclocked 175Mhz Core and 250Mhz Memory via MSI Afterburner. GPU still managed to boost itself up to about 1490Mhz sometimes 1500Mhz.

For Synthetic benchmarks…

In 3DMark Firestrike:

3DMark Score: 7162
Graphics Score: 8108
Physics Score: 12233
Combined Score: 2869
Test 1: 38.7 fps
Test 2: 32.37 fps
Physics Test: 38.84
Combined Test: 13.35

Please note that with Unigine tests, that MFAA was on, and also that with the Heaven benchmark, I recorded the entire test with Shadowplay, impacting performance to some degree. Framerates will be rounded up to nearest .5 decimal from here on.

Unigine Valley. Max Settings, 1080p, MSAA x8 with MFAA
Minimum: 19
Average: 36
Maximum: 68

Unigine Heaven Max Settings, 1080p, MSAA x8 with MFAA.
Minimum: 14
Average: 32
Maximum: 69

And now for some of the game benchmarks. All tests were done at 1080p and measured with FRAPS benchmarking tool. Bear in mind that these are guidelines as different configurations will have varying performances.

GTA V. Ultra settings, no advanced settings and MSAA off. Population options were also maxed out, tested with the in-game benchmark.
Minimum: 18
Average: 91
Maximum: 257
Do not be fooled. The maximum only occurred whilst loading the benchmark. I WILL say however that the game does run around 75-90FPS, at least with my system in normal gameplay. It is very playable, though GTA V isn’t the most demanding of games when you use modest settings. MSAA is the main killer of framerate in this game, so use at your own risk.

Battlefield 4. Ultra settings, MSAA x2 and MFAA on. Two tests, one with 100% Resolution Scale and 120% Resolution Scale.
100% Resolution Scale
Minimum: 49
Average: 72
Maximum: 112
120% Resolution Scale
Minimum: 47
Average: 58
Maximum: 67

Battlefield 4 isn’t as demanding as it used to be, at least with current gen hardware. Do note however that increasing the resolution scale doesn’t reflect performance as using DSR or using higher resolutions, and is a little finicky with performance. Nonetheless, 1080p in BF4 with this card is easily playable at Ultra settings, and more so when overclocked. In the 100% test, I almost never dipped below 60, staying around 65-70 in most scenarios. In the 120% test, it was more around the low to mid 50’s at best, but still, not bad at all! Since Battlefield 4 uses a similar engine to Star Wars Battlefront, performance should be around the same level, if not a little less. To compare, with my 750 Ti, playing at similar settings gave me about 40-45fps minus the MFAA, so there was a good jump there.

Killing Floor 2. Maxed out settings, no Nvidia Flex.
Minimum: 0
Average: 82
Maximum: 101
Killing Floor 2 is a great game to test this card out. While it’s not as crazy as such as Crysis 3 or something like Witcher 3 in terms demand, it is actually quite demanding when zeds and gore fill the screen excessively. Nonetheless, this card trumps it with more than playable framerate, though due to its early access nature, it does have very minor hitching, hence the 0 frames recorded on the minimum. It is improving fast however! Turning on Flex will impact performance, bear that in mind.

Metro Last Light. Maxed out settings, Advanced PhysX off, SSAA off. Tested via the Benchmark tool.
Minimum: 28
Average: 58
Maximum: 98

The benchmark tool is much more demanding than normal gameplay, but that’s to be expected. Turning on SSAA or PhysX in the benchmark and game will impact performance, but for the most part, in actual game I was able to play it without so much of a problem, sticking around the 60-70s. Very demanding sections can drop frames to lower numbers but never expect it to be a major problem. I do know however that Metro does prefer Nvidia cards by a slight amount, so it’d be difficult to give an apples to apples comparison, but in general with my 750 Ti, I was getting about 35fps Average.

Performance is all fine and dandy, running well with a lot of games right now. Take into account other benchmarks with other games as well, and with modest settings, you can still play your favourite Triple A type games without compromising heavily on fidelity or framerate.

Temperatures are also good despite it being a reference card. With 70% fan speed, I get an idle of 30C and on load, around 55-60C in Unigine tests and games. Fan also didn’t seem to whine heavily at this speed and was not very audible, if at all. Non-reference design fans will prove to be quieter and better at cooling however, as well as being quieter in general.

So, right now practically it’s the perfect mid-range card, right? Almost. The competing cards, the Radeon 380 and the Radeon 380X, offer better stock performance. The 380X especially outclasses both the 380 and 960, so it’d be easier to give an apples to apples comparison to the 960 to the 380 since the 380X is designed to be the gap filler between mid-range and high end.

However, the 380X is more expensive than both cards, but if you’re willing to throw in the extra £30, it can be worth it. The AMD cards however do not have the same overclocking potential as the 960, though the 960 HAS to be overclocked to have similar performance to its primary competitor, the Radeon 380, and the 380 can get a slight boost when that card is overclocked. Not everyone would be familiar with overclocking, and it can be risky in on itself, especially if you are tampering with the voltage. For the average PC user, they buy a card, stick it in and play games. In general, the 380 and 960 are very close to performance in most games when at their highest potential, and some games favouring one over the other can also be a factor.

So why would you buy the 960 when the 380 and 380X just offers more performance without overclocking?

I can think of several reasons:
1: Nvidia’s GeForce Experience and Shadowplay. Some people may not like proprietary software, but GFE is actually a good one in my books. It can schedule updates and you can check if there are any updates needed, and express install from GFE itself. It also scans for games, and notifies what features your system is capable of such as LED Visualizer and Shield, and Shadowplay. Shadowplay itself is quite lovely, allowing users to record gameplay footage, and stream directly to Twitch, with minimal impact to performance. It has a range of tools to help with such features as well, and one of my favourite features is Game Optimization, which allows users to optimize a game by changing several in-game settings to provide the best balance between performance and looks. This makes the 960 a little bit user friendly for less Computer Savvy gamers.

2: Small Form Factor 960s. Small form factor versions of the 380 exists, but it’s 4GB variant is marginally more expensive than a 960 4GB SFF. On Scan.co.uk, the 960’s cheapest ITX 4GB variant was £163, vs the 380’s ITX 4GB variant is £183, which is £20 more for a small gain in stock performance. Now for most users, a full sized 960 or 380 would suffice, but for users who are building HTPCs in small cases or users with small cases that can fit full size cards but want as much breathing room as possible, it would make sense to go with the ITX cards, and the 960’s ITX offerings are usually cheaper and perform similarly when tuned correctly. Some users have strangely designed cases that are height conscious, or length conscious. The 960 has a slew of different SFF based 960s, 2GB and 4GB variants, and do not trade any performance whatsoever. The card I have in question has a small PCB and only the fan makes it longer than it is. Most of the SFF 960s have above reference design clock speeds, and have varying voltage profiles, perhaps offering better overclocked performance than my reference 960 in the end. I am limited by height due to the design favouring side mounted HDDs. Previously I fitted a 660 Direct CUII, but was forced to remove one of the HDD mounting points to close the case due to the copper heat pipes making the card very tall. The reference 960 is just short enough to allow the case to close and allow me to fit back the HDD mount. Some 380s however will also not have that problem if you’re willing to go for the full sized card.

3: Power cables. Power consumption to me is something I don’t personally care about. What I’m more concerned about is whether I have enough cables, or the right cables, to fit my specific card in to work. For most enthusiasts, they would probably have a good PSU to handle any single card they can throw at them, but what about the budget builders and OEM system users with restrictive PSUs? My T5500 has two 6-Pin PCIe connectors, and can easily fit a full sized 380, but the 960 in most cases require only one 6-pin connector. You could use a molex adaptor, but such adaptors may not provide enough power for cards in general, especially when it comes to 6-pin to 8-pin adaptors, take that into account. For users with lower end PSUs or PSUs with lower wattages, they will be limited by the connectors they have. For the 380, it requires two 6-pin connectors as mentioned, whilst the ITX version requires an 8-pin (or a 6+2-pin) connector. Some 960s and 380s require 6-pin and 8-pin connectors, but all in all, you can easily fit a 960 to most systems that has 6-pin connectors with its PSU, whilst the 380 won’t be as easy to fit in budget minded builds or OEM systems that do not have the extra connectors to spare. Also consider that for my system, it is SLI or Crossfire compatible, and since I have an extra 6-pin, I can easily fit a second 960 down the line. Neato. Then again I could’ve bought a 970 and not have considered that possibility.

4? Performance with lower end CPUs. With recent tests with Digital Foundry, they have discovered that AMD isn’t as friendly with lower end CPUs like the i3, losing some of their performance more so than Green Team offerings. This is HIGHLY subjective however and may be due to system configuration more than anything.

Of course, I’m not saying to ignore the 380 or 380X. Actually, have a look at those cards. If nothing is holding you back, then just go for it! But if you prefer Green team, still, props to the decision in the end. All three cards are well placed in the performance spectrum, and if you really want the features more and don’t mind losing a few frames vs Red Team’s cards, the 960 is the perfect midrange card for you! If raw performance is more your game, the 380 or 380X are great cards to look at, but consider that the 380X is a little oddly placed as the 390 and 970 aren’t that far up in price but offer a substantial jump in performance.

For me, it was a big jump from a 750 Ti, and the 750 Ti was a jump for me when considering I have an 850M Laptop GPU.

The 960 4GB KFA2 Reference gets a Pa-Proval, and I give it a respectful 9/10.
In general, the 960 cards get 9/10
It’s a great mid-range card, and while Red Team has slightly better offerings in performance right now, the extra features, the use of few connectors and the slew of different variants offering extra advantages makes a 960 one of my favourite cards of all time. It misses the 10 score since it loses to the 380 and 380X in stock raw performance, but those cards came after the 960, meaning Red Team had an edge in fine tuning the competing cards. Overall, overclocking the 960 helps it compete more closely with the 380, so to really differentiate the two cards, you have to consider whether the extra GeForce based features are for you, or whether you prefer slightly better general performance.

The GTX 750 Ti was once a fairly popular card for mainstream PC gamers and those who aren’t interested in the high end segment of Graphics Cards. Since the release of the GTX 950 however, the 750 Ti was overshadowed by better overall performance, and when competing with the likes of the lower priced Radeon 260X (with close performance) and the similar priced Radeon 270 (which usually outperforms the 750 Ti) at the time of its release, it was a hard sell for some. At the same time, the 950’s announcement and release had Nvidia dropping the price of their once £120-130 card to ranges of £90-100, which it now competes with the likes of the Radeon 360 in terms of price and performance. But why would you want to buy a 750 Ti when the 950 and even the Radeon 370 offers better performance for a small price adjustment? There is only one reason you would buy a 750 Ti over any other budget card: low power consumption and most 750 Ti cards do not require any power connector, and relies only on the PCIe slot. 

But then we have this model, Gigabyte’s entry to the 750/750 Ti Low Profile market. Very few vendors make Low Profile cards, others include Zotac and Galaxy/KFa2, not to mention, MSI. This card is a Low Profile that takes up two slots, but it shouldn’t be a problem to fit this card in a number of extremely small ITX cases or OEM systems with a PCIe x16 slot. Gigabyte recommends a PSU wattage of 400w, however, the card’s TDP is only 60w, possibly this recommendation was a sort of ‘better safe than sorry’ situation.
This card costs similar or exactly the same as non-low profile cards, which would sport better coolers, some cards may require an extra 6-pin for better overclocking. So why would you want to buy a Low Profile over a standard card? It would make sense if you don’t have a lot of space in the case or that the system you’re going to slot it into only accepts LP cards.

But does this mean that this card’s smaller size mean lesser performance? Let’s find out! For the sake of performance comparison, I will be comparing the 750 Ti’s performance to its descendant, the 850M which shares its specifications but only uses DDR3 VRAM.

Test Setup: Dell OptiPlex 990 SFF
CPU: Intel Core i5 2400 Quad Core
RAM: 4x2GB DDR3 Single Channel setup
GPU: GTX 750 Ti 2GB GDDR5 (Gigabyte Low Profile OC)
HDD: 1TB Western Digital Caviar Black Enterprise
OS: Windows 10 Pro
PSU: 240W Gold Rated OptiPlex SFF Proprietary
Native Resolution: 1360x768 (DSR Factor: On, x1.2 Native)
Card Overclock: +155Mhz Core, +280Mhz Memory

Test Setup: Clevo W650SJ
CPU: Intel i5 4210M Dual Core Hyperthreaded
RAM: 2x4GB DDR3 Dual Channel
GPU: GTX 850M 2GB DDR3
HDD: Toshiba SSD 128GB, Seagate Laptop Thin SSHD 500GB
OS: Windows 10 Pro
Native Resolution: 1366x768
Card Overclock: +155 Mhz Core, +135Mhz Memory

Card Specifications:
Gigabyte 750 Ti 2GB Low Profile
Shading Units: 640
TMUs: 40
ROPs: 16
SMM Count: 5
GPU Clock: 1032 MHz
Boost Clock: 1110 MHz
Memory Clock: 1350 MHz, 5400 MHz effective
Memory Size: 2048 MB
Memory Type: GDDR5
Memory Bus: 128 bit
Bandwidth: 86.4 GB/s

GTX 850M DDR3 Model
Shading Units: 640
TMUs: 40
ROPs: 16
SMM Count: 5
GPU Clock: 902 MHz
Memory Clock: 1001 MHz, 2002 MHz effective
Memory Size: 2048 MB
Memory Type: DDR3
Memory Bus: 128 bit
Bandwidth: 32.0 GB/s

Considering that I was limited on options for my test system, this was the perfect system for testing two different theories: Whether OEM systems would be good for gaming and whether a SFF OEM system with a low wattage PSU would be able to provide for the 750 Ti even when boosted via Overclocking. Having extensively tested with this system in demanding games, I have not had any sort of power issue (yet) and I previously had the GT 545 which uses slightly more wattage (70w) and yet I still OC’d that card by 150Mhz Core. Outside of testing I can OC the card by 170Mhz Core and 350Mhz Memory, but sometimes the card boosts itself further due to GPU Boost 2.0.

Since the Dell system was connected to a HD Ready TV, playing at 1080p was possible but all in all the quality was poorly scaled and made games look bad. Instead, I decided to use Nvidia’s DSR options, adding x1.2 the native resolution, and in calculation found that it was close to 1440x900, which isn’t that far off from 900p (meaning negligible performance differences). Rather than stay conservative with my settings, I went as far as turning all settings in game to Ultra and adding some form of deferred Anti-Aliasing to give an idea of how it’d perform at allocated settings. However, 1080p tests will not be included for the purpose of this review.

To obtain the Minimum, Average and Maximum FPS of games, I use fraps FPS benchmark. I am also testing with the OC on at all times, to show the full potential of the card without any BIOS hacking or voltage modifications. Also, with the specifications of the system I had to hand I believe that the performance would be in line with similar lower tier Intel i5 CPUs, i3 CPUs and High tier AMD Quad Cores and AMD 6 Core or even AMD 8 Core CPU’s. For the most part, my system would be more representative of a low tier i5 and i3 system. RAM Setup however is mostly due to what I was able to work with, but RAM all in all makes mostly little difference in actual gameplay. All in all, the specifications are modest, perfect for something such as a 750 Ti.

Games Tested:
Battlefield 4 Multiplayer:
Ultra Settings x2 MSAA 1360x768 (DSR x1.2)

Killing Floor 2:
Ultra Settings 1360x768 (DSR x1.2)

Far Cry 4:
Ultra Settings SMAA (DSR x1.2)

Counter Strike Global Offensive:
Max Settings MSAA x2 1360x768 (DSR x1.2)

GTA V Benchmark:
High Settings Max Textures 1360x768 (DSR x1.2)

Metro Last Light Benchmark:
Very High Settings, SSAA and Advanced PhysX off 1360x768 (DSR x1.2)

Synthetic Benchmark: 3DMark

OptiPlex 990
3DMark Firestrike Default (Demo):
3DMark Score: 4342
Graphics Score: 4917
Physics Score: 5773
Combined Score: 1932
Graphics Test 1: 23.47 fps
Graphics Test 2: 19.63 fps
Physics Test: 18.33 fps
Combined Test: 8.99 fps

W650SJ
3DMark Firestrike Default (Demo):
3DMark Score: 3080
Graphics Score: 3649
Physics Score: 4282
Combined Score: 1190
Graphics Test 1: 17.67 fps
Graphics Test 2: 14.4 fps
Physics Test: 13.6 fps
Combined Test: 5.54 fps

The W650SJ fought valiantly, and despite the 850M sharing similar specifications to the 750 Ti, the DDR3 VRAM reduces its performance and overclocking potential is also limited. Also, add to the fact that it is a laptop card with less power being delivered and slight architectural differences, it couldn’t compete with its big brother. Pairing it with a Dual Core Mobile i5 also impeded the score especially in the Physics test and Combined test.

Killing Floor 2
Average FPS: 80
Minimum FPS: 49
Maximum FPS: 91
Verdict: The performance of the game is to be expected, despite being a major upgrade in visuals. The main stress comes from the overabundance of gore and giblets that fill the screen when tearing through hordes of zeds. All in all, more than playable, with minimum framerates near impossible to achieve without combining explosions after explosions.

Battlefield 4
Average FPS: 77
Minimum FPS: 48
Maximum FPS: 110
Verdict: In multiplayer, the map I chose was Parcel Storm which makes use of a lot of weather based particle effects. During the benchmark, it seemed that I’ve caught myself in an intense firefight, and also the match was a 64 Player match. The game looks great, but still found strong performance, with the minimum frame rate only occurring during really intense firefights filled with explosions, and I found myself reaching close to the Max FPS in game more times during clear moments. Minimum FPS was few and far between and never lasted more than a few seconds. If I’m honest, in most other maps you’d expect performance around the 80-90FPS mark.

Far Cry 4
Average FPS: 51
Minimum FPS: 28
Maximum FPS: 76
Verdict: Considerably a heavier game and with its heavy settings, I expected slightly lower performance. I was impressed, though I did notice that the minimum FPS only occurred during moments where the game just locked down to 30FPS. Strange to say the least, but the card handled FC4 at Ultra settings. Changing the settings down to High and turning off DSR can help improve performance. Indoors, the FPS would usually shoot closer towards the Max FPS.

Counter Strike Global Offensive:
Average FPS: 157
Minimum FPS: 107
Maximum FPS: 267
Verdict: You shouldn’t have a problem with this game, especially with the allocated settings. Playing this game at 1080p with the GPU still goes above 100+ FPS, so it is more than perfect for Counter Strike.

GTA V Benchmark
Average FPS: 82
Minimum FPS: 51
Maximum FPS: 112
Verdict: At High settings, GTA V is easy to run for most cards, perhaps even at 1080p you could run these settings above 40-50FPS. The minimum FPS only occurred during the last moment of the benchmark involving an explosive chain reaction. Expect slightly better performance in game, though tuning up the settings to Very High rather than High will have a noticeable impact on FPS performance.

Metro Last Light Benchmark
Average FPS: 50
Minimum FPS: 97
Maximum FPS: 17
Verdict: The test went through without a hitch, and ran fairly smoothly despite the crowded firefights and explosions and fire effects dousing the screen. The minimum framerate only occurred during dropped frames, otherwise, expect better performance in game for most moments.

Note: The GTAV and Metro Last Light Benchmark may not be representative of actual gameplay as they are designed to stress the system to a higher degree in comparison to standard gameplay.

Card Temperatures (Measured via MSI Afterburner)
Average Load: 70-75C (Dell) 60-70 (W650SJ)
Max Temperature: 80C (Dell) 75C (W650SJ)
Average Idle: 35-40
750 Ti GPU Fan Speed: 100%
W650SJ Fan: Full Speed

Conclusion: Despite the card’s performance at 1080p not being the greatest to satisfy the needs of gamers targeting 1080p60 at High/Ultra settings stably or at all, the 750 Ti is capable of delivering playable FPS at 1080p at High/Ultra settings for players who don’t mind the trade in smoothness for better visuals. However, in the case of players more than willing to sacrifice visuals or even scale down the resolution of their game to play, they will be surprised to see that the 750 Ti excels at performance at 900p and lower or at High/Medium or Medium settings for games. Games that are less demanding however are playable at 1080p or beyond at higher resolutions above 60FPS, such as Source games as such.

But for £30-50, you could get your hands on a GTX 950, and perhaps a little more can stretch you to a 960 which would be better for most users building their own PC and they offer superior performance, and are true 1080p cards. The 750 Ti can be considered a 1080p card but lose out in terms of raw performance, but the Gigabyte LP (and other LP 750/750 Ti cards) can be applied to a wider array of systems, especially OEM systems lacking 6-pins or ones that cannot fit standard size cards.

For those with lower resolution TVs and Monitors, or those who need a card to fit very small cases as well, this card is great as a Home Theatre/Home Entertainment graphics card, able to play games and improve multimedia experience. It can also sport up to four screens via three HDMI ports and a single display port.
In my personal opinion, my SFF OptiPlex 990, once a simple business/personal desktop is now an all-around entertainment system. With the performance at the resolution and settings I tested at, I was more than happy to play with the system that I have, and where I have applied it to makes it a very nice ‘LAN Box’, where I could in theory take this little PC to LAN parties or elsewhere with ease. In terms of the 750 Ti, I believe it is a card best played at either 1366x768/1360x768 for the highest in game settings and high frame rate, but 900p should deliver better resolution scaling for 1080p monitors and still allow the card to play at Ultra/High settings in most games at playable frame rate or frame rate above 60FPS, 1080p is where this card starts to strain as games would require lower settings or disabling AA to achieve 50-60FPS in higher end titles.

My verdict however is strong. Gigabytes card is able to handing very high clock speeds (even boosting beyond what I’d expected with my overclocks) and at the settings I played at, handled most games without a hitch.

Temperatures aren’t amazing but I tested it in a very small case with a system where I couldn’t change the fan speed myself due to the BIOS. All in all, it still handles itself well, and the small card does its job of keeping it cool. Despite being at 100% fan speed, it is not very audible during a typical gaming session or watching videos, and you’d have to be quite close to hear the whine. In a very quiet room, you may hear it, but I’ll be honest, it’s much quieter than my Clevo W650SJ. Be advised however, 2GB may be a strain for certain games at 1080p, otherwise it should be fine at slightly lower settings or resolutions.

Final Score: 9/10
Not a perfect card, but a great card. It provides performance and relative value, especially for its size. Small package but efficient performance. I doubt there would be another LP card to compete with its performance alone unless Nvidia or AMD announced a new PCIe based LP card that could match or outpace a 750 Ti. There may be other cards that can beat this card for performance, but none could fit so easily into ITX systems. A recommendation from me!

Hello people, I am assuming you are all sitting comfortably as you look at the screen. Good, now let me talk to you about this new budget GPU I bought for a mere £30 from CeX (Computer Entertainment Exchange). Ordered online and with a £2.50 Delivery charge, I bought this ole GPU for £32.50, in hopes of giving a Dell Optiplex 990 something to use so that it can play games. I couldn't afford a GTX 750 or 750 Ti, so I did my research.
Actually, I ordered the GT 440 OEM 1.5GB GPU, which is pretty much similar to the 545 but with lower clock speeds. Getting it in the mail, I installed it and found out it was the GT 545 instead. Because the Dell PC (Which I affectionately named 'Della') doesn't have any power connectors, I was limited to cards like this, that don't use power connectors and that are short in size. Though I can't close the case and I had to remove the I/O shield for the GPU to fit it, the system is running fine.
I decided to do a few tests so that I can see how this card performs in games, and Della is hooked up to our HD Ready TV which means I kept it at 1360x768, the best scaled resolution for this TV.

Specifications (Della)
Intel Core i5 2400 
4GB DDR3 1333Mhz RAM (2x2GB)
GT 545 3GB DDR3 (OC'd 850Mhz Core, 1040Mhz Memory)
320GB HDD
240W Gold Rated Proprietary Dell PSU
Windows 7 Home Basic

Counter Strike Global Offensive
    Max Settings, x2 MSAA, Vsync off
    Average FPS: 100+
    Verdict: A game like Counter Strike is easily playable with such a card like this, however, note that I wasn't able to achieve such frame rate at stock clocks. More or less, less demanding games are easily playable, especially at this resolution. More or less, you can crank up the details or resolution to your liking, and naturally this card is more than capable of handling it.

Killing Floor 2 Beta
    Max Settings, SSAO, Vsync off, Laboratory Level
    Average: 30FPS Average, 27FPS during intense battles and 35-40FPS in less demanding scenes
    Verdict: Though still in its beta stages, KF2 runs fairly well on a number of machines. What surprised me is that I was able to run the game at these settings with fair framerate at all. Take in to consideration that this card was released in May 2011, though the same specifications were found in the 440 OEM which was released in October 2010. Scaling down some of the settings is a must for smooth gameplay, unless 30FPS with drops is fine for your taste.

GTA V
    Very High Textures, High Shaders, Lowest Population settings, Ambient Occlusion Normal, Tesselation Normal, everything else on Normal settings
    Average: 35-40FPS average
    Normal (AO Off, Tesselation Off, Textures High): 50+ Average
    Verdict: GTA V is a very scalable game, and since the patches, is more easier to run with older or lower end machines. Despite the shortcomings of the GT 545, it is capable of playing this game without a hitch. Though you're not going to be playing at high or very high settings other than textures, but the game still looks fair enough to the eye and runs smoothly at its lowest settings, hitting 50-60FPS. Not bad for a four year old card.

Conclusion: I've yet to test out other games, and though my test mythology wasn't too deep, I'm more than certain what this card is capable of. It's a small card, taking only a single slot and armed with a small fan. For the most part, with the overclock, it'll hit 60-70C during load/gaming. It's best suited for Micro ATX or Micro-ITX based cases, unless you're willing to shave off a part of your case to fit the card at all.
Would I recommend this over a GT 640, GT 740 or a Radeon 240, Radeon 250 or any other PCIe powered cards? That depends on whether you plan to buy them new or second hand. If you can find any of these cards within the £30-40 price range, you will not be disappointed if you're only looking to fit it in for the while. I paid only £30 for my card, but a lot of people would be peeved by the DDR3 memory, but let me add that the bandwidth is helped by the 192bit bus, and with my OC, it hits 50GB/s, whether it'll be able to utilize the whole 3GB of VRAM, it's still faster than a lot of the DDR3 equipped cards on the market. The GT 545 and GT 440 OEM were mostly found as OEM cards, hence their cheap price but also rarity. More or less, modern PCIe powered cards may beat this card at the settings I've played the games at, but buying certain cards like the GT 630 for £40-50 new is absurd, as most cards below the X40 mark (for Nvidia and AMD) are usually heavily cut down.
Overall, this card is a great quick fix, allowing a user to play games with reduced eye candy. Its a cheap card, but still for the price it was for, a card that might be for you if you want to get in to PC gaming through baby steps. My 850M is still better, but I like this card for what it provides at its price point. If you can find one (or a 440 OEM) at £30-40 second hand, you might want to try it out. Though there are times where more powerful cards can be found as cheap, but this card does not need extra power connectors, allowing it to fit in a lot of computers. My Dell PC only has 240W, but it provides sufficient power nonetheless.

So would I recommend it? Yush :3