GTX 770 4GB vs 2GB Showdown

This very specific GTX 770 4GB versus GTX 770 2GB comparison is a follow up to our EVGA GTX 770 4GB review last week. It attempts to demonstrate the performance differences between a 4GB GTX 770 and a GTX 770 with 2GB at resolutions of 1920×1080, 2560×1600 and 5760×1080. We are going to compare two GTX 770’s clocked the same, and identical in every respect, except that one is equipped with 2GB whereas the other has 4GB of GDDR5.  Above are pictured the EVGA GTX 770 that comes equipped with 4GB and the reference GTX 770 that comes standard equipped with 2GB. Both GTX 770s use the same GPU, and when clocks are set similarly, the performance is effectively identical as long as the framebuffer of the 2GB card is not exceeded.

EVGA GTX 770 4GB vs Nvidia GTX 770 2GB benchmarked

This particular evaluation is designed to compare 4GB vRAM versus 2GB.  Since we do not want any chance of our CPU “bottlenecking” our graphics, we are testing our two graphics cards by using our Ivy Bridge Intel Core i7-3770K at 4.50GHz, 16GB Kingston “Beast” HyperX DDR3 at 2133MHz, and an EVGA Z77 FTW motherboard. The Core i7-3770K at 4.5GHz is more than enough to differentiate even high-end video cards at high resolution and at high detail settings.

The EVGA Z77 FTW motherboard features 16x + 16x PCIe 3.0 specification for CrossFire/SLI which we will test in the next article covering GTX 770 SLI as we have just received a 2GB GTX 770 Hall of Fame edition from Galaxy for evaluation. Unfortunately,  we were unable to SLI our 4GB EVGA GTX 770 with the 2GB reference GTX 770 even using Coolbits, and we did not want to use older (modded) drivers.

The EVGA GTX 770 SC uses the same PCB as the reference version, but we note that it uses double capacity GDDR5 modules totaling 4GB instead of the standard 2GB. There is one interesting thing that we did not cover in our original evaluation of the EVGA GTX 770 SC.  It has a dual-BIOS switch.  We can’t wait to test out two GTX 770s against each other, but before we begin the testing, head over to our testing configuration.

Test Configuration – Hardware

• Intel Core i7-3770K reference 3.50 GHz/Turbo to 3.7GHz, overclocked to 4.5 GHz; HyperThreading is on, supplied by Intel.
• EVGA Z77 FTW motherboard (latest Beta BIOS, USB/PCIe 3.0 specification; CrossFire/SLI 16x+16x), supplied by EVGA
• 16 GB DDR3 HyperX Kingston “Beast” RAM (2×8 GB, dual-channel at 2133MHz; supplied by Kingston)
• EVGA GTX 770 SC 4GB (factory overclocked to 1111/7010MHz but clocked to Boost to 1162MHz) on loan from EVGA.
• Nvidia GTX 770, 2 GB (reference base clocks of 1046/7010MHz, overclocked to Boost to 1162MHz), supplied by Nvidia
• Onboard Realtek Audio
• Genius SP-D150 speakers, supplied by Genius.
•  500 GB Seagate Barracuda 7200.10 hard drive
• 240GB HyperX SSD, supplied by Kingston
• Cooler Master Platinum Pro 1000W PSU, supplied by Cooler Master
• Thermaltake Overseer RX-I full tower case, supplied by Thermaltake
• Philips DVD SATA writer
• HP LP3065 2560×1600 thirty inch LCD.

Test Configuration – Software

• NVIDIA GeForce 326.19. High Quality; Single-display & multiple-display Performance mode; Prefer Maximum Performance
• Windows 7 64-bit; very latest updates
• EVGA PrecisionX 4.21 overclocking and monitoring software
• Latest DirectX
• All games are patched to their latest versions.
• vsync is forced off in the control panels.
• Varying AA enabled as noted in games; all in-game settings are specified with 16xAF always applied; 16xAF forced in control panel for Crysis.
• All results show average frame rates
• Highest quality sound (stereo) used in all games.
• Windows 7 64, all DX9 titles were run under DX9 render paths, DX10 titles were run under DX10 render paths and DX11 titles under DX11 render paths.

The Benchmarks

• Synthetic

  • 3DMark 11
  • 3D Mark 2013 – Firestrike/Firestrike Extreme
  • Heaven 4.0
  • Valley 1.0

  DX9

  • Serious Sam 3 BFE
  • Alan Wake: Ameican Nightmare
  • The Witcher 2
  • Borderlands 2
  • Aliens: Colonial Marines
  DX10
  • Crysis
  DX11
  • BattleForge
  • Alien vs. Predator
  • STALKER, Call of Pripyat
  • Metro 2033
  • Civilization V
  • Lost Planet 2
  • Total War: Shogun II
  • Crysis 2
  • Dirt 3
  • DiRT: Showdown
  • Batman: Arkham City
  • the Secret World
  • Battlefield 3
  • Max Payne 3
  • Sleeping Dogs
  • Sniper Elite V2
  • Far Cry 3
  • Hitman: Absolution
  • Assassin’s Creed III
  • Crysis 3
  • Tomb Raider 2013
  • BioShock: Infinite
  • GRID 2
  • Metro: Last Light

Using EVGA’s PrecisionX

Before we get to the performance charts, let’s see how we equalized the GTX 770 2GB reference version and the EVGA 770 SC 4GB version clocks for identical performance. We used the EVGA tool, PrecisionX, to set the clockspeeds of each card to equalize performance.

Here are the specifications for the reference GTX 770:  The reference GTX 770 is clocked at 1046MHz with a guaranteed Boost to 1085MHz.  In the case of our own reference GTX 770, it boosts to 1136MHz with the Power and Temperature sliders maxed out. In the case of the EVGA card, it is clocked at 1111MHz or +65MHz higher than the reference GTX 770.

However, things are still not equal as the EVGA card will boost to 1162MHz.  So we downclocked the EVGA card to 1046MHz base clock, and overclocked the reference version +26MHz so the boosts were identical at 1162MHz.   Since our cards never throttled, the performance became effectively identical between the two cards with the amount of vRAM being the only potentially differentiating factor.

Let’s head to the performance charts to see how the now performance-equalized EVGA GTX 770 SC 4GB card compares with the reference GTX 770 2GB card when we attempt to stress the framebuffers with maximum settings, high AA, and super-widescreen resolutions.

Performance summary charts & graphs

Here are the summary charts of 30 modern PC games and 4 synthetic benches. The highest settings are generally chosen and it is DX11 when there is a choice; DX10 is picked above DX9, and the settings are generally ultra or maxed unless specified on the chart. Specific settings are listed on the charts. The benches are run at 1920×1080, 2560×1600 and 5760×1080. “Wins” are in Bold.  If the scores are identical both results are in bold.There isn’t a lot of difference between the cards at 1920×1080 or at 2560×1600.  We only start to see minimal differences at 5760×1080, and even so, there is rarely a frame or two difference.  If we start to add even more AA, in most cases, the frame rates will drop to unplayable on both cards.

Let’s only look at the games where there is more than a single FPS difference between the two GTX 770s.Metro: 2033 is completely unplayable on either card at our highest resolution, and even GTX 770 4GB SLI wouldn’t be playable either.  Sleeping Dogs has a problem actually displaying on the outer LCDs although the performance is cut, so this benchmark has to be discounted.  This leaves five games out of 30 where a 4GB GTX 770 gives more than a 1 frame per second difference over the 2GB-equipped GTX 770.  And one of them, Metro: Last Light still isn’t even quite a single frame difference.

Of those five games, two of them are unplayable at 5760×1080 although in these cases, 4GB GTX 770 SLI would finally make some sense over 2GB GTX 770 SLI.  That only leaves Lost Planet 2 and two racing games that gain some advantage by choosing a single GTX 770 4GB card over the single GTX 770 2GB.  And in Lost Planet 2, we were able to add even higher anti-aliasing – from 8xAA to CSAA8XQ and to CSAA32X – but the performance difference was greatest with 8xAA.

There is one last thing to note with Max Payne 3:  It would not normally allow one to set 4xAA at 5760×1080 with any 2GB card as it claims to require 2750MB.  However, when we replaced the 4GB GTX 770 with the 2GB version, the game allowed the setting.  And there were no slowdowns, stuttering, nor any performance differences that we could find between the two GTX 770s.

Let’s head for our conclusion.

Conclusion

This has been quite an enjoyable exploration for us in comparing the 4GB vRAM-equipped EVGA GTX 770 SC to the 2GB reference GTX 770 at the same clocks.  Unless a gamer plays at 5760×1080, we wouldn’t recommend choosing 4GB vRAM over 2GB for a single GTX 770 for today’s games.  However, if a gamer is planning to SLI GTX 770s at 5760×1080, then it might be reasonable to pick the 4GB version.  We can’t predict future performance of upcoming games so we will leave them out of our equation.

Pros – 4GB over 2GB

• The EVGA GTX 770 4GB SC  provides excellent performance for a 4GB $459 card even when clocked to the reference card’s clocks and it provides marginal improvements at super-widescreen resolutions.
• SLIing two or more 4GB cards has more room for better performance at 5760×1080 over SLI’ing 2GB cards.

Cons

• Cost.  4GB-equipped GTX 770s cost about $40 to $50 more than their 2GB counterparts.
• 4GB is generally unnecessary for 1920×1080 or even for 2560×1600.  Only marginal benefits are found at 5760×1080 for a single card.

The Verdict:

• We would not recommend buying a single GTX 770 4GB card over the 2GB version if there is a significant price difference.  We might recommend choosing 4GB GTX 770 SLI over 2GB GTX 770 SLI, but only if the gamer plays at super-widescreen resolutions or requires extreme anti-aliasing.

Stay tuned, there is a lot coming from us at ABT.  Next week, expect an evaluation of Cooler Master’s Comforter Air laptop cooler.  We also received a Hall of Fame GTX 770 from Galaxy and there will be a showdown with the EVGA GTX 770 SC as well as a comparison with the reference GTX 770, also scheduled for next week.  And it will lead to a GTX 770 SLI evaluation the week after. Please don’t forget to check our Forums. ABT forum tech discussions are among the best to be found anywhere.  Happy Gaming!