780 ti in modern games. Video cards. Conclusions on synthetic tests

NVIDIA has decided to introduce another video card based on the GK110 chip, most likely the last one. This time GeForce GTX 780 Ti is already in use full version Kepler architecture, which will continue to dominate the market until Maxwell arrives next year. Judging by numerous rumors, the idea of ​​introducing such a video card was born at NVIDIA quite a long time ago. Ever since the advent of the corresponding Tesla and Quadro video cards, it became clear that a desktop video card is just around the corner. The new GeForce GTX 780 Ti video card can be seen as an answer to AMD Radeon R9 290X, but in any case, this is NVIDIA's new flagship. The GeForce GTX Titan video card gives way to the GeForce GTX 780 Ti. But how much faster will the new product work compared to Titan? Or compared to Radeon R9 290X? In our article we will answer all these questions.

But before we move on to the final chapter in Kepler's life, let me introduce small excursion to the past. On March 22, 2012, NVIDIA introduced the GeForce GTX 680 video card, the first model based on the new “Kepler” GPU architecture. The streaming multiprocessors known from the "Fermi" architecture were improved in GK104 from 32 to 192 CUDA cores, now called SMX clusters. However, the changes affected not only an increase in the number of cores, but also the proportions of control logic and computing units - in the new generation, the emphasis was shifted to computing units. Eight SMX clusters of 192 CUDA cores each resulted in a total of 1536 cores.

For almost a year, NVIDIA continued to optimize the "Kepler" architecture. As a result, in February 2013 the GK110 processor was released, containing up to 15 SMX clusters with 192 CUDA cores, which gave 2880 cores. But for the GeForce GTX Titan and the subsequent GTX 780, fewer SMX clusters were used: 14 and 12, respectively. Here, it seems to us, the reason was the difficulties of chip production, since a crystal with 7.1 billion transistors and an area of ​​533 mm² is not so easy to produce without errors using a 28-nm process technology. Even during the presentation of the first desktop video cards based on the GK110 chip, rumors began to circulate about the possible release of “GeForce GTX Titan Ultra”. And today this video card finally appeared.

Before we move on to a detailed look at the GeForce GTX 780 Ti, let us talk about the specifications.

NVIDIA GeForce GTX 780 Ti
Retail price RUB 24,990 in Russia
649 euros in Europe
Products webpage NVIDIA
Technical Specifications
GPU GK110 (GK110-425-B1)
Technical process 28 nm
Number of transistors 7.1 billion
876 MHz
928 MHz
Memory frequency 1.750 MHz
Memory type GDDR5
Memory capacity 3.072 MB
Memory bus width 384 bit
336 GB/s
DirectX version 11.1
Stream processors 2880
Texture blocks 240
48
Pixel fill rate 42 Gpixel/s
SLI/CrossFire SLI
TDP 250 W

Of course, the distinguishing feature of the GeForce GTX 780 Ti is the GK110 processor in its full version. It is equipped with 15 SMX clusters, each with 192 CUDA cores, for a total of 2880 CUDA cores. The graphics processor is manufactured using a 28 nm process technology and contains 7.1 billion transistors in an area of ​​533 mm². NVIDIA, unlike the latest "Hawaii" video cards from AMD, decided to set a fixed base GPU frequency of 876 MHz, but thanks to GPU Boost, dynamic overclocking is provided to at least 928 MHz. According to NVIDIA, most video cards can easily reach the 1 GHz bar in practice. We get a second NVIDIA graphics card with a high GDDR5 memory frequency of 1750 MHz. It is connected via a 384-bit interface, resulting in a memory bandwidth of 336 GB/s. But NVIDIA decided not to standardly equip the video card with 6 GB of video memory; unlike the GeForce GTX Titan, you will get only 3 GB. Perhaps the situation will change after the release of the first video cards with alternative cooling systems, so GeForce GTX 780 Ti with 6 GB of memory and even more may appear on the market. After equipping the Radeon R9 290X and 290 4 GB of memory, such a step seems quite logical to us.

The new GPU is equipped with 15 SMX clusters, which corresponds to 2880 CUDA stream processors and 240 texture units (16 of them are used per SMX cluster). The 384-bit memory interface interfaces with 48 raster operation pipelines (ROPs). According to the manufacturer's specifications, the theoretical pixel fill rate is no less than 42 Gpixels/s. In the SLI configuration, up to four GeForce GTX 780 Ti video cards can operate. Unlike high-end AMD video cards, the SLI bridge is still required for the new NVIDIA video cards to work. But the situation may change in the next generation. The thermal package (TDP) is the same 250 W - this may be another advantage over the Radeon R9 290X, whose TDP exceeds 250 W.

Screenshot of GPU-Z video card NVIDIA GeForce GTX 780 Ti

The GPU-Z screenshot confirms the key technical specifications and provides some additional details.

Comparison of NVIDIA GeForce GTX 780 Ti with competitors
Model AMD Radeon R9 290X NVIDIA GeForce GTX 780 NVIDIA GeForce GTX 780 Ti NVIDIA GeForce GTX Titan
Retail price about 480 euros 20.2 thousand rubles
about 410 euros
24,990 rubles
649 euros
32.2 thousand rubles
about 820 euros
Products webpage AMD NVIDIA NVIDIA NVIDIA
Technical Specifications
GPU Hawaii XT GK110 (GK110-300-A1) GK110 (GK110-425-B1) GK110 (GK110-400-A1)
Technical process 28 nm 28 nm 28 nm 28 nm
Number of transistors 6.2 billion 7.1 billion 7.1 billion 7.1 billion
GPU clock speed (base frequency) - 864 MHz 876 MHz 837 MHz
GPU clock speed (Boost frequency) 1.000 MHz 902 MHz 928 MHz 876 MHz
Memory frequency 1.250 MHz 1.502 MHz 1.750 MHz 1.502 MHz
Memory type GDDR5 GDDR5 GDDR5 GDDR5
Memory 4.096 MB 3.072 MB 3.072 MB 6.144 MB
Memory bus width 512 bit 384 bit 384 bit 384 bit
Memory Bandwidth 320.0 GB/s 288.4 GB/s 336 GB/s 288.4 GB/s
DirectX version 11.2 11.1 11.1 11.1
Stream processors 2.816 2.304 2.880 2.688
Texture blocks 176 192 240 224
Raster Operation Pipelines (ROPs) 64 48 48 48
TDP > 250 W 250 W 250 W 250 W

Comparisons with competitors from both our own production and AMD indicate the theoretical advantages and disadvantages of the GeForce GTX 780 Ti. The number of stream processors does not exceed its competitors so significantly; NVIDIA's memory bandwidth is slightly higher. But the memory capacity, Boost frequency and pixel fill rate (due to the higher number of raster operation pipelines) are higher for AMD. However, PowerTune and GPU Boost technologies today have a significant impact on performance, which is difficult to estimate based on theoretical specifications. So comparison of theoretical computing performance, memory bandwidth and other technical characteristics It’s hardly that relevant. The test results will show how far the new video card can break away from the GeForce GTX Titan and Radeon R9 290X.

Hi all. In this review, as the happy owner of an Asus Geforce Gtx780 Ti, I will try to reveal its main advantages and disadvantages, but I will start with a preface.

What's bad about 2x Asus Matrix HD7970?

What prompted me to do this was not the mania to spend money, but the deplorable situation of AMD in the field of drivers. The fact is that if one Asus Matrix worked perfectly, then with the second the situation changed radically. For example:

When using Catalyst 13.11 Beta, the second video card was loaded up to 99% when idle, and warmed up well. But the first one was fine. And if this was solved by rolling back the drivers to 13.4, then I lost about 20% in performance. And the problem was precisely in the drivers. Yes, and stability suffered, FPS drops also did not suit me. But this is not the only reason. I was not entirely satisfied with the home “stove” in the literal sense, although I placed 10 turntables in the Corsair Carbide Series 400R case. And they consumed a lot.

With the GTX780 Ti the situation is different: the performance is higher (with more stable operation), the heating is less (especially since the air was thrown out), and the “green” ones have much better firewood. Also included is Geforce Experience with ShadowPlay, game optimization, driver control, and the ability to change the visualization of the “GEFORCE GTX” backlight on video cards with reference CO. But more on all this later.

Since I sold my Asus Matrix on the 27th, I had to think quickly. On the one hand, I could take the version from Gigabyte with their WindForce, but the warranty and availability decided everything. On Sunday I went and bought, perhaps, one of the few Gtx780 Ti brought to Sochi.

Design and Cooling System

Perhaps I'll start with this. Now all the latest reference SBs from Nvidia look almost the same, below is a photo comparing the Gtx780 Ti, Gtx Titan and Gtx780

Well, the internal component

And also characteristics

I probably won’t write much about the effectiveness of the Nvidia reference system. If, like me, you are not very bothered by noise (maybe you are wearing headphones), or simply do not pay attention, then Msi Afterburner’s own profile will help you, and thanks to this, even with overclocking, the video card does not heat up more than 80*C even in the heaviest applications. But I think that dropsy would not hurt such a device. Fortunately, it's coming Asus Geforce GTX780 Ti ROG Poseidon. And Asus Gtx780 Ti with DirectCU II is also just around the corner

Or you can install Arctic Cooling Accelero Hybrid.

Yes, by the way, the air does not stagnate in the housing - the turbine is maximum speed(more than 4200Rpm) blows air no worse than any 120 fan at a speed of 2000Rpm. The only downside is the noise.

Performance and overclocking

Application Performance

As befits a high-end product, the video card provides excellent performance in any application. And, as mentioned above, it does this stably, without drops in FPS. Crysis 3 also turns out to be quite playable at maximum settings. However, nothing else can be expected. It should also be noted that during idle (or under light load) the frequency drops to 324Mhz, and under light load - approximately 700Mhz.

In idle time

Below are videos from Battlefield 4, Crysis 3, Saints Row IV and Far Cry 3 Blood Dragon. Maximum settings.

Tests were carried out on the configuration:

Kingston HyperX 8Gb 1600Mhz

OCZ Vector 128Gb

Seagate Constellation 3Tb 7000Rpm, 128Mb

Battlefield 4

Crysis 3

Far Cry 3 Blood Dragon

Saints Row IV

And also temperature indicators

Battlefield 4

Crysis 3

Far Cry 3 Blood Dragon

Saints Row IV

Overclocking

The Gtx780 Ti copes well with overclocking, frequencies of 1080 for the core and 1838 for memory were excellent, there are no artifacts or FPS drops, the voltage is also stable. And thanks to Boost, the frequency increases to 1132. But this is the case, according to Gpu-Z. In applications I observed 1209Mhz. Compared to the factory overclock (about 928 in Boost and 1000Mhz in reality), this is a good option. The only sad thing is that the power reserve is 106%. But it's not that simple. If you believe the official data, the TDP of the GTX780 Ti is 250W, and with an increase in power reserve, these 106% are counted not from 250, but from 300W! In the end, everything turns out very well. And yes, the video card with a custom CO profile does not heat up above 80*C, the frequency is not reset, and we have smooth gameplay in any games.

Screenshot of Gpu-Z at standard frequencies

Screenshot of Gpu-Z overclocked

If you want to overclock your video card even more, then you need to resort to unlocking the voltage for Msi Afterburner, everything is described below.

You should start by checking the PWM controller for compatibility; to do this, enter in the command line: msiafterburner /ri4,20,99 or msiafterburner /ri3,20,99. If the program returns the value 41, then you can move on. In the folder with MSI Afterburner, go to the profiles directory. We find the file with the card ID and add at the end:

VDDC_Generic_Detection=0

VDDC_NCP4206_Detection=4:20h

VDDC_Generic_Detection=0

VDDC_NCP4206_Detection=3:20h

GeForce Experience

It is worth mentioning separately about the new application from Nvidia, Geforce Experience.

1.) Here you have game optimization, which works in most licensed games (list on the Nvidia website), you don’t have to configure anything. Although sometimes there are mistakes - for example, I found 3x Crysis 3, 2x Fallout 3, 2x Skyrim, and 2x Borderlands 2.

2.) Checking and automatic (or at the user’s request) installation of drivers, including Beta versions.

3.) ShadowPlay. This is a separate utility from Nvidia that uses a separate H.264 codec to record video from games without significant performance loss. Of course there are losses, but within 10 Fps. This does not affect gameplay. You can select the recording type (background and manual; background; manual; broadcast on Twich;). In background mode, you can select the recording duration, maximum 20 minutes. with a size of 7.5GB. It should also be noted that recording is only in 1080p.

4.) Backlight visualizer. Works on reference Gtx690, 770, 780, Titan and 780 Ti. There are many modes, I won’t describe them.

My opinion

When I bought this device for 30,000, I was hoping to get maximum performance, and I got it. Yes, it was possible to take a Gtx780 with Direct CU II, but I didn’t want to. And the warranty here is 3 years, in any case, this device will be relevant throughout this and next year.

If you are an avid gamer or enthusiast, then you should definitely pay attention to this card. Excellent build quality, optimized drivers from Nvidia, maximum performance + excellent overclocking potential. Premium product, nothing to add.

Pros and Cons

On to the pros It's worth mentioning the enormous performance - the most powerful single-chip video card! More good drivers and a new Geforce Experience application with useful features

To the disadvantages It is worth including only the reference cooling system and the starting price in DNS

The notorious competition between the two main GPU manufacturers NVIDIA and AMD has once again given green fans cause for joy. Before the Reds had time to enjoy the applause in honor of the release of their new flagship in the form of the Radeon R9 290X, the Californians cleverly tripped them up. It was quite expected for experts that after the release of a top-end video card from AMD, NVIDIA would not stand aside and would try to create, if not much more, then certainly no less powerful solution. Expectations were justified, and a new representative of the GeForce family is being released - a video card GTX 780 Ti.

The announced video adapter at the end of 2013 is the most powerful among single-chip ones in terms of implementation in modern demanding games. The video card is built on a graphics processor labeled GK110, which was previously found in and. However, unlike the mentioned predecessors, the new product has a fully functional (not cropped) core, which is found only in a professional solution like . So, for example, in the GeForce GTX 780 Ti the number of computing cores is 2880, while in TITAN there are 2688 of them. But let's take a closer look at the characteristics of the above video cards in order to compare them.

Specifications

As can be seen from the table, the new product is ahead of its predecessors in many key parameters. Titan only has more video memory, but we already wrote about how important this parameter is in the article about. Thus, the final performance of the GTX 780 Ti is, if not superior, then at least significantly higher than that of the GTX 780 and GTX TITAN. Well, now, actually, about performance.

Synthetic test results

*Highest possible quality with screen resolution 1920x1080

And in conclusion of the review, I would like to say that the recommended price for the GeForce GTX 780 Ti video card for the US market is $699, for Russia - 24,990 rubles. Arrival of new items on Russian market expected after November 15, 2013.

Comparative GeForce testing GTX 780Ti and AMD Radeon R9 290X

advertising

Let's remember the line of NVIDIA video cards. GeForce GTX 770, GTX 780, GTX Titan and GTX 690 – a slightly incorrect, but understandable classification model range company accelerators. But GTX 780 Ti? For what? Where? What to compare it with? The answer is not as simple as one might think at first glance. If you think about it, then...

The sixth GeForce series was very clear: each next-ranking model was distinguished by a large number in its name. And now video cards appear with an index that includes 7x0. At the same time, the GTX 690 has not disappeared anywhere; it still remains the fastest dual-core solution. Maybe the GTX Titan is in the wrong group? Quite, why not, since it came to the gaming market from the world of computing and remains the ultimate offering for both gaming and computing.

The question arises: should NVIDIA discontinue the GTX 780 Ti after releasing it? The answer is also simple. For what? It is popular among those people who need computing and enthusiasts, however, for gaming best choice should be the GTX 780 Ti. This is because users have finally been provided with a fully functional GPU with 2880 stream processors. Yes, only now, after certain changes, which will be discussed below, the GK110 graphics processor is ready to join gaming applications and show what it is capable of.

Probably everyone is wondering why GK110 received a new name, or rather stepping B1, and not, say, A2? There is an opinion that numbers indicate corrections in metal connections inside the GPU. The letters indicate changes in the transistors themselves. In any case, there are no radical changes inside the crystal - it’s still the same GK110. All NVIDIA had to do was adapt the full version of the GK110 to the required heat dissipation at the calculated frequencies, and this was not easy to do.

The secrecy of the company's developments is covered in such a veil of secrecy that it is almost impossible to get intelligible answers; you will simply be bombarded with advertising or general phrases, from which it is impossible to find out the technical data. Well, secrets must be kept undisclosed, even under pointed questions from the press. For our part, we can only guess what tricks we had to go to in order to fit the new GPU revision into the allotted power consumption range.

advertising

Technical features

As you already understood, there are no physical changes in the main logical circuits. It is possible that optimizations have appeared internally that reduce the time it takes for jobs to complete within the GPU, which results in fewer transistors working simultaneously. This operating scheme has existed for a long time and is called “dark silicon”. No system can cool all 7.1 billion transistors operating at the same time, which means we must constantly monitor the balance, on the one side of which is performance, and on the other, frequency, power consumption and the resulting temperature. The better and more economically the gates work, and the lower the temperature is maintained, the faster the calculations occur.

Even before the advent of Hawaii as a competitor, NVIDIA introduced several provisions for GPU operation. So, the base frequency is the lowest level of GPU operation, GPU Boost is the average GPU frequency in games. More often than not, even after the video card was under a gaming load for a long time, GPU Boost kept the GPU frequency slightly higher than stated. AMD has taken a different path - the only mode in which the video card fully realizes its power is “Normal” or, as it is also called, Uber mode.

But unlike its competitor, NVIDIA believes that the user does not need to deal with either BIOS switches or driver settings - the video card will do everything for them, and on its own. What we can really blame the GeForce developer for is that the fixed power consumption limit is too close to the factory setting. It remains to be seen whether the engineers have really solved the general problem of a sharp increase in energy consumption?

Specifications

NameR9 290R9 290XGTX 690GTX 780GTX 780 TiGTX Titan
CodenameHawaiiHawaiiGK104GK110GK110GK110
Technical process, nm 28 28 28 28 28 28
Size of core/cores, mm 2 438 438 294x2 521 521 521
Number of transistors, million 6200 6200 3540x2 7100 7100 7100
Core frequency, MHzUp to 950Up to 1000 915 (1020) 860 (900) 880 (930) 840 (880)
Number of shaders (PS), pcs. 2560 2816 3072 2304 2880 2688
Number of rasterization blocks (ROP), pcs. 64 64 64 48 48 48
Number of texture units (TMU), pcs. 160 176 256 192 240 224
Maximum fill rate, Gpix/s 60.6 64 58.6 41.4 42 40.2
Maximum texture sampling speed, Gtex/s 151.5 176 234.2 165.7 210.2 187.5
Pixel/vertex shader version 5.0 / 5.0 5.0 / 5.0 5.0 / 5.0 5.0 / 5.0 5.0 / 5.0 5.0 / 5.0
Memory typeGDDR5GDDR5GDDR5GDDR5GDDR5GDDR5
Effective memory frequency, MHz 5000 5000 6000 6000 7000 6000
Memory capacity, MB 4096 4096 2048x2 3072 3072 6144
Memory bus, bit 512 512 256x2 384 384 384
Memory bandwidth, GB/s 320 320 192x2 288.4 336 288.4
Power consumption (2D / 3D), Wnd / ndnd / ndnd / 300nd / 250nd / ndnd / 250
CrossFire/SliYesYesYesYesYesYes
Recommended price at the time of announcement, $ 399 549 999 499 699 999

Appearance and dimensions

  • Analysis of geometric average results, purchase attractiveness and measurement of energy consumption
  • advertising

    Introduction

    After the release of AMD's flagship single-processor solutions - Radeon R9 290X and Radeon R9 290, NVIDIA lost its leadership in the gaming video card market, as its GeForce GTX Titan and GTX 780 accelerators were inferior to direct competitors.

    However, the company did not put up with this state of affairs, and in a fairly short time, NVIDIA released its response to the opponent’s actions - the GeForce GTX 780 Ti 3072 MB video card. Why is it remarkable and thanks to what hidden resources Will the new product be able to compete with opposing models?

    Firstly, it is based on a full-fledged and uncut GK 110 graphics processor, which includes 2880 unified shader processors, 240 texture units and 48 raster operation units. The GPU itself operates at 876 MHz.

    Secondly, the effective operating frequency of the video memory was 7000 MHz, which, together with the 384-bit bus, made it possible to increase the video memory bandwidth to 336 GB/s. This was enough to neutralize the 512-bit memory bus Radeon video cards R9 290X, the video memory bandwidth of which is 320 GB/s.

    advertising

    In this test we will study what the new NVIDIA product is capable of.

    The rivals of the GeForce GTX 780 Ti 3072 MB are:

    • GeForce GTX Titan 6144 MB;
    • GeForce GTX 780 3072 MB;
    • GeForce GTX 770 2048 MB;
    • GeForce GTX 680 2048 MB;

    • Radeon R9 290X 4096 MB;
    • Radeon R9 290 4096 MB;
    • Radeon R9 280X 3072 MB.

    Test configuration

    Tests were carried out on the following stand:

    • CPU: Intel Core i7-3770K (Ivy Bridge, D2, L3 8 MB), 1.0 V, Turbo Boost / Hyper Threading - off - 3500 @ 4600 MHz (1.25 V);
    • Motherboard: GigaByte GA-Z77X-UD5H, LGA 1155, BIOS F14;
    • CPU cooling system: Corsair Hydro Series H100 (~1300 rpm);
    • RAM: 2 x 4096 MB DDR3 Geil BLACK DRAGON GB38GB2133C10ADC (Spec: 2133 MHz / 10-11-11-30-1t / 1.5 V), X.M.P. - off;
    • Disk subsystem: 64 GB, SSD ADATA SX900;
    • Power unit: Thermaltake Toughpower 1200 Watt (standard fan: 140 mm inlet);
    • Frame: open test bench;
    • Monitor: 27" ASUS PB278Q BK (Wide LCD, 2560x1440 / 60 Hz).

    Video cards:

    • Radeon R9 290X 4096 MB - 1000/5000 @ 1130/5800 MHz (Sapphire);
    • Radeon R9 290 4096 MB - 947/5000 @ 1120/5800 MHz (Sapphire);
    • Radeon R9 280X 3072 MB - 1000/6000 @ 1150/7000 MHz (Gigabyte);

    • GeForce GTX 780 Ti 3072 MB - 876/7000 @ 1110/7700 MHz (MSI);
    • GeForce GTX Titan 6144 MB - 837/6008 @ 970/7200 MHz (Gigabyte);
    • GeForce GTX 780 3072 MB - 863/6008 @ 1000/7200 MHz (Palit);
    • GeForce GTX 770 2048 MB - 1046/7000 @ 1260/7800 MHz (Zotac);

    • GeForce GTX 680 2048 MB - 1006/6008 @ 1260/7100 MHz (Gainward).

    Software:

    • Operating system: Windows 7 x64 SP1;
    • Video card drivers: NVIDIA GeForce 334.67 Beta and AMD Catalyst 14.1 BETA 1.6.
    • Utilities: FRAPS 3.5.9 Build 15586, AutoHotkey v1.0.48.05, MSI Afterburner 3.0.0 Beta 18.

    Testing tools and methodology

    For a more clear comparison of video cards, all games used as test applications were launched in resolutions of 1920x1080 and 2560x1440.

    Built-in benchmarks, FRAPS 3.5.9 Build 15586 and AutoHotkey v1.0.48.05 utilities were used as performance measurement tools. List of gaming applications:

    • Assassin's Creed 4 Black Flag(Nassau).
    • Batman: Arkham Origins (Gotham City).
    • Battlefield 4 (Tashgar).
    • Company of Heroes 2 (Benchmark).
    • Crysis (Benchmark - Village).
    • Far Cry 3 (Chapter 2. Hunter).
    • GRID 2 (Benchmark).
    • Max Payne 3 (Chapter 5. Alive, although a little shabby).
    • Metro: Last Light (Benchmark).
    • Saints Row IV (Start of the game).
    • Sleeping Dogs (Benchmark).
    • Tom Clancy's Splinter Cell: Blacklist (Item zero).

    Measured in all games minimum And average FPS values. In tests in which there was no possibility to measure minimum FPS, this value was measured by the FRAPS utility. VSync was disabled during testing.

    Let's move directly to the tests.