All graphics cards that are tested in the new 3DMark
2013 in a system with the following specifications:
The motherboard
·
Intel Siler DX79SI (Intel X79 Express, LGA 2011,
BIOS 0494
CPU
·
Intel Core i7-3960X Extreme Edition, 3.3 GHz,
1.2 V, 6 x 256 KB L2, 15 MB L3 (Sandy Bridge-E, C1, 32 nm)
CPU cooler
·
Phanteks PH-TC14PE (fans 2 x 135 mm at 900 RPM)
Temperature Interface
·
ARCTIC MX-4
Graphics Card on AMD graphics processors
·
HIS 7970 IceQ X2 GHz Edition 3 GB (H797QMC3G2M)
1050/6000 MHz;
·
HIS 7950 IceQ X2 Boost Clock 3 GB (H795QMC3G2M)
950/5000 MHz
·
Sapphire Radeon HD 7870 GHz Edition OC 2 GB
1050/5000 MHz
·
HIS 7850 IceQ Turbo X 2 GB (H785QT2G2M)
·
AMD Radeon HD 7770 GHz Edition 1 GB 1000/4500
MHz
·
AMD Radeon HD 7750 1 GB 800/4500 MHz
Graphics Card on Nvidia graphics processors
·
NVIDIA GeForce GTX 690 2x2 GB 915/1020/6008 MHz
·
ASUS GeForce GTX 680 DirectCU II TOP 2 GB
(GTX680-DC2T-2GD5) 1137/1202/6008 MHz
·
Gigabyte GeForce GTX 670 Ultra Durable 2 GB
(GV-N670OC-2GD) 980/1059/6008 MHz
·
ASUS GeForce GTX 660 Ti DirectCU II TOP 2 GB
(TI-DC2T-2GD5) 1059/1137/6008 MHz
·
ASUS GeForce GTX 660 DirectCU II OC 2 GB
(GTX660-DC2O-2GD5) 1020/1085/6008 MHz
·
Gigabyte GeForce GTX 650 Ti 2 GB (GV-N65TOC-2GI)
1033/5400 MHz
System Memory
·
DDR3 4 x 4GB Mushkin Redline (Spec: 2133 MHz /
9-11-10-28 / 1.65 V)
System Disk
·
Crucial m4 256 GB SSD (SATA-III,CT256M4SSD2, BIOS
v0009)
Disk for apps and games
·
Western Digital VelociRaptor (300GB, SATA-II,
10000 RPM, 16MB cache, NCQ) in cool and more silent Scythe Quiet Drive 3.5” HDD
Back-up disk
·
Samsung Ecogreen F4 HD204UI (SATA-II, 2 TB, 5400
RPM, 32 MB, NCQ)
System Case
·
Antec Twelve Hundred (front panel: three
Noiseblocker NB-Multiframe S-Series MF12-S2 fans at 1020 RPM; rear panel: two
Noiseblocker NB-BlackSilentPRO PL-1 fans at 1020 RPM; upper panel: standard
200mm fans at 400 RPM)
Control and Supervision Panel
·
Zalman ZM-MFC3
Power source
·
Seasonic SS-1000XP Active PFC F3 1000 W (with a
standard 120 mm fan)
Display
·
27” Samsung S27A850D (DVI-I, 2,560x1,440, 60 Hz)
So, we have 6
graphics accelerations on AMD GPUs and other six on the Nvidia GPUs. The
"red" side is represented by the products from three manufacturers:
HIS, Sapphire and AMD as the following:
The
"red" side: HIS, Sapphire and AMD
The glory of the "green" side will
be protected by six graphics cards from three different manufacturers: Nvidia,
Asus and Gigabyte:
The
"green" side: Nvidia, Asus and Gigabyte
As you may have noted, 10 out of 12 graphics
has increased the default frequency. However, we decided not to lower them to
the nominal reference value, because the different performance will be very
clear, and besides, most of the graphics cards being sold are those with the speeds
that have been increased slightly. Therefore, we will test all products
"as they were".
To lower the dependence of graphics card
performance on the overall platform speed, I overclocked our 32nm six-core CPU
with the 37x multiplier, the BCLK frequency at 125MHz and enabled the
"Load-Line Calibration" up to 4,625GHz. The Vcore Processor is boosted
to 1.49 V in the BIOS motherboard:
Overclocking
the 32nm six-core CPU with the 37x multiplier
The Hyper-Threading Technology was enabled. The
DDR3 system memory of 16GB operates at 2GHz frequency with the period of
11-10-28 and 1.65V voltage.
All the tests were done in Microsoft Windows 7
Ultimate x64 SP1 with all critical updates installed since that day and the
following drivers:
·
Intel Chipset Drivers 9.4.0.1014 WHQL for the motherboard
chip
·
Library of DirectX End-User Runtimes
·
AMD Catalyst 13.2 Beta 5 Drivers (12.100.17.0) since
5/2/2013 + Catalyst Application Profiles 12.11 (CAP2) for those cards based on
AMD graphics
·
Nvidia GeForce 313.96 Beta Drivers since 29/01/2013
for graphics cards based on Nvidia
All tests were run
with the standard installation settings. We did not adjust the default settings
when launching them in any way. Take a look at the observed results.
Using
only one CPU core
Our testing operated at the same rate
regardless of the types of cards based on AMD that we installed. Nvidia-based
cards made outputs more volatile. Measurement standards developers actually
warned that this could happen because the load would be too slow for modern
desktops.
