iXBT Labs - Computer Hardware in Detail






AMD 790GX Chipset

Powerful integrated graphics and CrossFireX support.

November 5, 2008

<< Previous page

     Next page >>

As always, we'll start with computational tests in order to find out how the integrated graphics core affects memory performance (Athlon 64 X2 4000+).

Test Int. graphics Discrete graphics (Radeon 1900XTX)
Gigabyte MA78GM-S2H (AMD 780G) ASUS M3N-HT Deluxe (nForce 780a) Foxconn A7DA (AMD 790GX) Gigabyte MA78GM-S2H (AMD 780G) ASUS M3N-HT Deluxe (nForce 780a) Foxconn A7DA (AMD 790GX)
RMMA, average read rate, MB/s 3384 3327 3402 3484 3414 3458
RMMA, maximum read rate, MB/s 6388 5640 6330 6568 6561 6534
RMMA, average write rate, MB/s 2738 2430 2298 2851 2620 2821
RMMA, maximum write rate, MB/s 5507 5721 5449 5779 5767 5750
RMMA, pseudo-random seek time, ns 32.8 32.7 32.1 31.9 32.2 32.1
Archiving with WinRAR, min:sec 2:30 2:33 2:31 2:28 2:32 2:30
MPEG4 (XviD) encoding, min:sec 6:04 6:01 6:02 6:04 6:00 6:00

There is nothing interesting in these test results. And it's good, because it means that performance corresponds to the CPU/RAM combo used in these tests.

Chipset and CPU Intel G35 NVIDIA nForce 780a AMD 780G
(500/1066 MHz)
AMD 780G
(overclocked to 950/1333 MHz)
(700/1333 MHz)
(700/1333 MHz)
Pentium E2160 Athlon X2 4000+ Athlon X2 4000+ Athlon X2 4000+ Athlon X2 4000+ Phenom X4 9550
Avg. fps Min. fps Avg. fps Min. fps Avg. fps Min. fps Avg. fps Min. fps Avg. fps Min. fps Avg. fps Min. fps
Company of Heroes (800x600), fps 12.5 4.0 23.6 14.5 38.6 21.5 40.6 20.0 60.6 24.6 71.1 27.0
Company of Heroes (1024x768), fps 9.7 4.0 18.1 10.2 28.1 13.8 40.7 22.6 43.3 17.8 48.2 17.3
S.T.A.L.K.E.R. (800x600), fps 28 9 62 16 68.0 22.5 91.9 24.2 84.8 20.1 97.2 30.4
S.T.A.L.K.E.R. (1024x768), fps 21 8 45 13 46.0 18.2 63.2 18.2 57.7 19.1 67.6 27.1
F.E.A.R. (800x600), fps 16 7 32 19 36 20 43 24 42 24 50 28
F.E.A.R. (1024x768), fps 11 5 22 12 24 13 29 15 27 15 34 19
World in Conflict (800x600), fps 14 9 32 13 39 14 41 14 43 13 45 18
World in Conflict (1024x768), fps 11 6 26 11 32 13 37 13 38 10 37 14
Serious Sam II (800x600), fps -   29   42.3   56.0   51.6   59.3  
Serious Sam II (1024x768), fps -   21   30.0   40.5   36.6   42.2  

Surprise, surprise! AMD 790GX performance is not just higher than that of 780G at the nominal frequencies. It generally surpasses results of the integrated graphics core in Gigabyte MA78GPM-DS2H overclocked to 950 MHz (the fastest modification of 780G with DDR3-1333 video buffer). Judging by results of the first motherboards tested, the 790GX itself also overclocks well. In particular, we managed to overclock the integrated core of the motherboard from Foxconn to the same 950 MHz, voltage raised by 0.18 V. And it's not the limit, as chipset cooling systems of 790GX-based motherboards are generally more powerful than on 780G products. By the way, performance of the overclocked 790GX is also impressive. It hardly makes sense for games, which frame rate already exceeds 50 fps (that is in such cases you can just raise quality settings). But performance in World in Conflict, which has grown to 62 fps at lower resolutions and to 52 fps at higher resolutions, shows that the game is worth the candle. Note that it makes no sense to install a weak processor on such a system -- you'll have to overclock it as well, lest the graphics core potential is not revealed completely.

It's rather difficult to compare with competing products here, as AMD apparently enjoys an imposing advantage. From the marketing point of view, this advantage may seem even too big and irrational -- why play all trump cards, if the opponent has nothing in response? But first of all, this is probably the way the manufacturer streamlines elements of the graphics core, which we'll see in promising Low-End models of AMD Fusion processors. Secondly, a noticeable advantage on the graphics side makes the entire AMD platform more attractive than an equally-priced motherboard+processor kit from Intel.

Unfortunately, we had no opportunity to run full tests of the motherboard on Intel G45, which had been delayed so much and failed to deliver what had been promised. From what we already know, this chipset outperforms its predecessor (G35) by 50% at best, which is totally insufficient to compete even with GeForce 8200. Manufacturers themselves are not inclined to provide such motherboards for tests. Most reviews just compare such motherboards with each other or with G35-based products. Besides, prices for already available motherboards are higher than even those for richly rigged up products based on GeForce 8200 and AMD 780G. It would have been funny, if motherboards on AMD 790GX had also been cheaper than G45-based ones. But frankly speaking, it would have meant that the market exists by unreal rules, and real product properties do not matter.

The situation with nForce 780a is slightly different. Motherboards on this chipset are rigged just like products on AMD 790GX, but they are designed for the absolutely different audience, those who want to build 3-Way SLI, nothing less. Such tough users are apparently not interested in the integrated graphics core. And this chipset is equipped with the integrated graphics system to support HybridPower in the first place. And NVIDIA engineers are of the opinion, that it won't be used in games. It's high time to take a look at power consumption in practice, as 790GX does not offer such technology, which is available in mobile chipsets from AMD. It features only generic functions to reduce average power consumption -- quite effective, and as we already found out in our ECS A790GXM-A tests, they help reduce idle power consumption even relative to motherboards on the economic 780G chipset).

Configuration (Athlon 64 X2 4000+ Windsor CPU, Cool'n'Quiet disabled) Mode Power consumption (entire testbed, W)
Idle 3DMark 06
nForce 780a+GeForce 9800 GTX Power Saving 69-93 -
nForce 780a+GeForce 9800 GTX Boost Performance 117-134 170-240
AMD 790GX+Radeon HD 4850   85-114 138-195

Radeon HD 4850 is the closest counterpart to GeForce 9800 GTX as far as performance is concerned. We used this chipset once to test HybridPower, so it's correct to compare such pairs. As we can see, the NVIDIA tandem generally consumes 20 W less in idle mode or when the integrated graphics core is used. But is it worth the inconveniences? HybridPower from NVIDIA is available only in Windows Vista. Users have to switch between modes manually (before and after gaming), video output is available only to monitors plugged to outputs on the motherboard. And 3D performance of a graphics card is 3-5% as low as when monitors are plugged to a graphics card directly. When we speak of a single graphics card, the answer can be only negative. Especially as the AMD system can "catch up" during a game owing to lower power consumption of Radeon HD 4850. Consequently, AMD's approach to power saving (perfecting PowerPlay technology, which helps the card consume less power both in idle and loaded modes) looks more thorough and convenient for users.

However, in case of 3-Way SLI, power saving grows proportionally, when HybridPower disables three cards (the 9800 GTX card is not the most voracious candidate for this role). And an opportunity to make the system friendlier to environment (and to users, if we take into account reduction of overall heat release and consequently of fan noise) can make up for the above-mentioned drawbacks.

Does it mean that everything is peachy and there is nothing to improve? On the contrary, we'd like to see these technologies develop to a friendlier and all-purpose implementation of these modes from both manufacturers. In the very least, the system should switch automatically to the integrated graphics core and allow to plug monitors not only to a motherboard, but also to a master graphics card (putting the other SLI-cards to sleep) at user's choice. The same functionality should be provided for desktop chipsets from AMD. For example, in case of CrossFire with two Radeon HD 4870 X2 cards, disabling these cards in idle mode will certainly save power.

What concerns HD Video decode tests, we already registered minimal load on motherboards with AMD 780G (read the details in our reviews of AMD 780 and NVIDIA nForce 700a). The 790GX chipset decreases this load by several percents more. However, the difference is not critical, so compact inexpensive motherboards on 780G remain an optimal choice for media centers, combined with Low-End models of Phenom. And if minimal power consumption is a must, it's hard to find anything better than Athlon X2 4850e with TDP=45 W.


AMD engineers demonstrated one more time that they can make their graphics cores even stronger (and faster than competing solutions). No doubts, the expensive purchase of ATI was the right move after all. Chipsets with integrated graphics are presently a serious argument for the AMD platform.

As for AMD 790GX, motherboards on this chipset are designed for a wide audience. On the one hand, gamers who like the modern series of discrete graphics cards from AMD, and who want an opportunity to install a couple of them in CrossFire mode eventually without paying much for this feature. Besides, motherboards with Socket AM2+ are compatible with future processors for Socket AM3. Considering 140-W voltage regulators on motherboards with 790GX, users will have plenty of opportunities to overclock such processors. In fact, it's wrong to generalize here, we should test and compare actual motherboards on this chipset, which will be especially interesting, because we get Advanced Clock Calibration along with other tools. So these motherboards will help users maintain their computers fit for modern games for a long time at moderate expenses.

On the other hand, such motherboards will also serve well those users who don't play games very often (or choose "quiet" genres, which depend less on the frame rate than dynamic shooters), and who are ready to use integrated graphics only, if its performance is at least on the level of cheap representatives of discrete cards. Owing to the new Southbridge and requirements to PCB design, motherboards with this chipset can be used in all-purpose easily upgradeable computers, equal in functionality with computers based on classic discrete chipsets without integrated graphics.

Write a comment below. No registration needed!

<< Previous page

Article navigation:

Page 1: Introduction, testbed

Page 2: Test results, conclusions

blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook

Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.