AMD 790GX: RV610 For Enthusiasts?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
[citation][nom]zenmaster[/nom]Nice Review, However I would like to see how well SLI scales on this board using a 4850 and 4870 vs another AMD Board with full Duel 16x PCIE 2.0 Slots.[/citation]

SLI won't scale at all. This board doesn't have SLI. Are you sure SLI means what you think it means...?
 
[citation][nom]zenmaster[/nom]Nice Review, However I would like to see how well SLI scales on this board using a 4850 and 4870 vs another AMD Board with full Duel 16x PCIE 2.0 Slots.I've seen some other reviews where Dual 8x PCIE 2.0 Slots start having scaling issues on the 4850 at are not seen on the 3800 Series.[/citation]

It's possible Zen, but I'd expect no more than a few percent drop-off as a result of the x8 slots. Perhaps the x16 vs x18 debate warrants a follow-up?
 
[citation][nom]computerfarmer[/nom]Thank you for this excellent article. Does the 128mb side-port limit the on board video to that amount of memory? Does this affect Vista Areo? [/citation]
Nope, you still have access to unified memory. The side-port simply helps boost performance until the core needs to go out to system memory, is my understanding.
 
[citation][nom]cshorte[/nom]good review,i have a question why do you recommend ddr1066 over 800? i thought there wasn't a huge difference.also im a bit confused about this 16x, vs. 8x 8x...if i use a single card (for now) which motherboard (790g, 790gx) will produce more graphixs capabilities?[/citation]

DDR2-1066 gives you more bandwidth, which is a major factor in the performance of the integrated graphics.

If you use a single card, that card will get 16 lanes of PCI Express connectivity. Not more capabilities, but more throughput to the card itself. With two cards, each card gets eight lanes instead, cutting back on bandwidth but increasing performance due to the scaling of a pair of cards in there.
 
[citation][nom]ltcommander_data[/nom]It would have been interesting to see a comparison between the GMA X4500 and the 790GX.http://www.hkepc.com/?id=1510&page=5&fs=idn#viewThe GMA X4500 seems to be much faster than the GMA X3500 and within 10-30% of the 780G in actual games. I guess the 790GX was a timely addition and with it being 20% faster than the 780G, that should open up the lead over the GMA X4500 to 30-50%.[/citation]

This is something we had planned, and will address in a forthcoming piece on integrated graphics chipsets, likely next month. We simply ran out of time to get G45 in there.
 
[citation][nom]kenyee[/nom]and which of these upcoming 790GX motherboards are microATX?The only one I know of is the DFI one...[/citation]

I have to imagine all of the majors will launch a microATX version at some point. I haven't seen the DFI board yet, but would certainly like to see what they're able to do with 790GX
 
slightly off topic...
will there ever be a 16x 16x to use both cards to their fullest capabilities?

can you illaborate on what bandwidth is for me? its been a while since i built a computer and im starting to do research now for my next one 😛
 
[citation][nom]snarfies1[/nom]SLI won't scale at all. This board doesn't have SLI. Are you sure SLI means what you think it means...?[/citation]

Think he meant CrossFireX. =)
 
[citation][nom]cshorte[/nom]slightly off topic...will there ever be a 16x 16x to use both cards to their fullest capabilities?can you illaborate on what bandwidth is for me? its been a while since i built a computer and im starting to do research now for my next one [/citation]

Yup, if you grab a 790FX-based board and drop in two cards, each will get 16 lanes of PCI Express 2.0. Bandwidth is the capacity a given pipeline has for moving data. With a x8 connection, the pipe can theoretically move 8 GB/s of information and a x16 link can move 16 GB/s (in both directions).
 
but u mentioned that the gx boards only performed slightly lower than? with SB750 put onto new FX boards can I expect something totally different?

sorry for the silly questions
 
[citation][nom]cshorte[/nom]but u mentioned that the gx boards only performed slightly lower than? with SB750 put onto new FX boards can I expect something totally different?sorry for the silly questions[/citation]

With FX you're going to be shedding the integrated core and getting support for as many as FOUR cards in CrossFireX mode (that northbridge has more PCI Express lanes available to it). The performance difference won't be massive, but the 790FX is definitely a better enthusiast platform.
 
Call me a stickler for detail, but I am a bit confused. A 1x PCIe 1x connection is capable of 2.5Gbit/s bandwidth. 2.5Gbit/s is roughly 320MB/s
(megabytes). The PCIe 2.0 standard is supposed to double this to 5Gbit/s, or 640MB/s. Now according to my math 8 x 640 would mean 5120MB/s or 5GB/s(gigabytes/second). Now since the PCIe 'bus' is unidirectional, and that PCIe can transmit this data in both directions at the same time this would be roughly 10GB/s *IF* you were to somehow max the data in both directions. This of course *is* theoretical, and likely that you will never see these numbers because of data overhead etc.

Anyhow if I am wrong, I am wrong, but I think at least I need some clarification. Maybe 8GB/s(including both directions) is accounting for data overhead already ? I do not know but 20% data overhead seems to be fairly high for this sort of data transmission.
 
Shouldnt they make budget boards (micro atx or pico atx factor) with just the 790gx integrated graphics?? SHould go well with X2 and 9150s
 
[citation][nom]yyrkoon[/nom]Call me a stickler for detail, but I am a bit confused. A 1x PCIe 1x connection is capable of 2.5Gbit/s bandwidth. 2.5Gbit/s is roughly 320MB/s(megabytes). The PCIe 2.0 standard is supposed to double this to 5Gbit/s, or 640MB/s. Now according to my math 8 x 640 would mean 5120MB/s or 5GB/s(gigabytes/second). Now since the PCIe 'bus' is unidirectional, and that PCIe can transmit this data in both directions at the same time this would be roughly 10GB/s *IF* you were to somehow max the data in both directions. This of course *is* theoretical, and likely that you will never see these numbers because of data overhead etc.Anyhow if I am wrong, I am wrong, but I think at least I need some clarification. Maybe 8GB/s(including both directions) is accounting for data overhead already ? I do not know but 20% data overhead seems to be fairly high for this sort of data transmission. [/citation]

In all actuality, you're very close here! Blatantly borrowed from Dell's well-written introduction to PCI Express:

"PCI Express bandwidth is commonly expressed as "encoded" bandwidth. PCI Express uses 8b/10b encoding, which encodes 8-bit data bytes into 10-bit transmission characters. This approach improves the physical signal so that bit synchronization is easier, design of receivers and transmitters is simplified, error detection is improved, and control characters can be distinguished from data characters.

The "encoded" bandwidth of a basic x1 PCI Express lane is 5 Gbps. However, a more accurate bandwidth figure is the "unencoded" bandwidth, which is 80 percent of 5 Gbps or 4 Gbps."

At the end of the day, you do lose 20% or so.
 
Thanks for the good review,

i am tempted to sell my DFI Lp DK M2RS and get this as apposed to buying a raid card.

I would have liked to see some crossfire scaling w/ 4850/4870's... would be interesting to see.

Once again thanks for the review and keep up the good work :)
 
[citation][nom]cangelini[/nom]It's possible Zen, but I'd expect no more than a few percent drop-off as a result of the x8 slots. Perhaps the x16 vs x18 debate warrants a follow-up?[/citation]

That is what I would expect as well, but I was analyzing some P45 Chipset reviews that tested CF scaling. Those boards have 8X PCIE 2.0 Slots.

There was little difference between the X48 Boards with Dual 16.0 PCIE 2.0 with the 3870 cards. However a Review that used the 4850s noted a 15% drop in performance.

Normally I would not expect the 4850 is saturate a 16x PCIE 1.0 Slow so I don't see why 8x PCIE 2.0 would have shown any issues.

The AMD Board may not have the same issue.
Or Perhaps the only Review that I saw of CF Scaling with the P45 Chipset using 4850s was flawed.
 
[citation][nom]cangelini[/nom]In all actuality, you're very close here! Blatantly borrowed from Dell's well-written introduction to PCI Express:"PCI Express bandwidth is commonly expressed as "encoded" bandwidth. PCI Express uses 8b/10b encoding, which encodes 8-bit data bytes into 10-bit transmission characters. This approach improves the physical signal so that bit synchronization is easier, design of receivers and transmitters is simplified, error detection is improved, and control characters can be distinguished from data characters.The "encoded" bandwidth of a basic x1 PCI Express lane is 5 Gbps. However, a more accurate bandwidth figure is the "unencoded" bandwidth, which is 80 percent of 5 Gbps or 4 Gbps."At the end of the day, you do lose 20% or so.[/citation]

Ah ha! Ok that makes more sense then. Still, now I am wondering if that includes data transmission overhead or not. Or even if there *is* overhead similar to networking data transmission overhead or not.
 
[citation][nom]Rhysee[/nom]Love the chick video .. obviously not a Tomshardware employee .. AMD prob?[/citation]

These two videos were created for Tom's by a channel publication. We do our own video in-house as well (check out the Vigor preview on the front page).
 
"i am tempted to sell my DFI Lp DK M2RS and get this as apposed to buying a raid card."

I'd personally wait until the 790FX + SB750 boards started to come out, as that would be a more proper upgrade from a 790FX-M2RS; keep the top-end northbridge, but get a better southbridge, instead of 'downgrade' to an IGP for the sake of a better southbridge. I'm personally waiting on the 45nm AMD chips, though, before I make any major system build plans.
 
Could you elaborate why ATI PowerPlay doesn't work? For me, the reason to get a powerful IGP is so I can run it under 2D and low end 3D to saving power and still have enough reserve if I want to play extreme games.
 
It isn't PowerPlay that we had trouble with. Rather, the functionality that AMD first explained when it launched 780G, whereby it'd turn off a discrete card if your display output came from the motherboard in a Hybrid CrossFireX mode.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom