Phenom ganged and unganged modes

[Tony]

Just some info for you guys.

ganged mode is 1x128bit dual channel

unganged is 2x64 bit dual channel

Some motherboards are reporting unganged as single channel at boot, this is not true, if you see anyone saying they have defective memory please point them to this post.

Ganged will give higher sandra bandwidth at this time than unganged, this is because sandra does not fully support phenom and is working in 1 thread mode. Once sandra supports Phenom unganged 2x64bit will take the lead

If you are running multi threaded benchmarks you need to be in unganged mode, this is the fastest way of running the system.

[KeZzZu]

Thanks for info, anyone know whats the difference between 790x and fx? except 2x more pcie slots

same overdrive ? setting possibilities etc?

[eva2000]

thanks for the heads up and info

[Lightman]

Good to know this from reliable source!

As of 790X chipset it will work with AMD Overdrive .

[macci]

Here is an example regarding Unganged performance and how the amount of test threads affects the bandwidth:



edit: just to clarify: Sandra uses 2 threads in case of K8 Dual Core CPU (the highest reference score in the screenshot)

[Tony]

Im going to push all the motherboard vendors I talk with to reword unganged single channel to dual channel 2x64bit mode on the boot screen, this simple misrepresentation is going to cause no end of RMA requests for all the memory manufacturers.

[Jamesrt2004]

Im going to push all the motherboard vendors I talk with to reword unganged single channel to dual channel 2x64bit mode on the boot screen, this simple misrepresentation is going to cause no end of RMA requests for all the memory manufacturers.

Wooaa You have POWER

but yea thanks for heads up as yea I would of been left scratching my head

[informal]

Im going to push all the motherboard vendors I talk with to reword unganged single channel to dual channel 2x64bit mode on the boot screen, this simple misrepresentation is going to cause no end of RMA requests for all the memory manufacturers.

Very good initiative ,will save a lot of time and money.

[gondezz]

Wow.. very valuable info. Thanks, Sir!
Can we choose between those options in BIOS? Or, is it done automatically?

[Tony]

you can choose

[mrcmtl]

whats the difference between 1 thread and 2 thread sandra test?

and does anyone know why phenom memory bandwidth is so low?

[justapost]

Im going to push all the motherboard vendors I talk with to reword unganged single channel to dual channel 2x64bit mode on the boot screen, this simple misrepresentation is going to cause no end of RMA requests for all the memory manufacturers.

Or

single dual channel 1x128bit for ganged and
dual single channel 2x64bit for unganged mode

AMD's K10 spec says there is a new channel interleaving option beside chip/bank interleaving.
Can channel interleaving be activated via the bios and does it have an impact on performance?

[Quintero]

AMD's K10 spec says there is a new channel interleaving option beside chip/bank interleaving.
Can channel interleaving be activated via the bios and does it have an impact on performance?

I think channel interleave is the same as unganged mode.

[Quintero]

Can we choose between those options in BIOS? Or, is it done automatically?

It's a BIOS setting, as Tony pointed out. But it should be noted that unganged mode will only be available on AM2+ boards.

[justapost]

I think channel interleave is the same as unganged mode.

Look at "F2x110 DRAM Controller Select Low Register" in AMD's BIOS and Kernel Developer’s Guide pg. 201.

4 DctGangEn: DRAM controller ganging enable. Read-write. 1=Both DCTs are ganged to form a single double-width DDR interface. 0=The DCTs operate independently. This also affects how DCT configuration registers; see section 2.8.1 [DCT Configuration Registers] on page 61. Note, if ganging is to be enabled, this bit must be set prior to programming any DCT registers.

2 DctSelIntLvEn: DRAM controller interleave enable. Read-write. 1=Channel interleave is enabled; DctSelIntLvAddr specifies which address bit is used to select between DCT0 and DCT1; this applies from the base system memory address of the node (specified by [The DRAM Base/Limit Registers] F1x[1, 0][7C:40]) to DctSelBaseAddr (if enabled). If the amount of memory connected to each of the DCTs is different, then channel interleaving may be supported across the address range that includes both DCTs, the top of which is specified by DctSelBaseAddr; the remainder of the address space, above DctSelBaseAddr, would then be allocated to only the DCT connected to the larger amount of memory, specified by DctSelHi.

It are different things and they can be configured independent.

[Quintero]

Ok, but I'm quite sure that unganged mode was referred to as channel interleave in some of the old previews.

DctGangEn definitely looks like the basic ganged/unganged option.

Not quite sure how DctSelIntLvEn works, but it seems that it sortof ping-pongs the addresses between the channels. However, I'd guess that could be done under the ordinary unganged mode as well. K10 is said to have better support for mismatched DIMMs, which seems to be at least part of what this feature does:

"If the amount of memory connected to each of the DCTs is different, then channel interleaving may be supported across the address range that includes both DCTs"

They also mention the word "node", which could mean that this feature applies to multisocket systems. But I'm not sure.

[Cooper]

So is there any difference on 780X what memory slots to use for "dual channel" ?

[Quintero]

So is there any difference on 780X what memory slots to use for "dual channel" ?

We'll probably see different solutions, just like with AM2 boards. Either way it shouldn't be chipset dependant.

On some boards, slot 1 and 2 constitute one channel, a design which favors scalability with four DIMMs rather than two. That's because the strobe mismatch between each slot within each channel would be smaller. The strobe delays are set on a channel basis, which means that each slot cannot be configured independently (each pair must share the same settings). Trace lengths and routing mismatches affect the strobe delays, and the smaller the distance between the slots that are sharing the same timings, the smaller the actual delay differences between the slots.

When using two sticks in dualchannel, the tables turn. Two sticks would normally scale a little bit higher on a board where slot 1 and 2 connect to different channels. Shorter traces mean reduced capacitive loading and possibly lower delays (like Max Async Latency), so with two slots populated it would normally be best to use the slots that are closest to the CPU..

[billb]

We'll probably see different solutions, just like with AM2 boards. Either way it shouldn't be chipset dependant.

That's because the strobe mismatch between each slot within each channel would be smaller. T.

ASUS A8N32's have adjustable "AI Skew" in the BIOS to compensate for trace lengths. A very useful tuning tool.

[Quintero]

ASUS A8N32's have adjustable "AI Skew" in the BIOS to compensate for trace lengths. A very useful tuning tool.

I have two great Asus AM2 boards myself (Crosshair and M2N32 WS PRO), which also have adjustable strobe delays and AI Skew. But I always leave them at auto, because I figure it's best to let the system calibrate the delays automatically during each initialization. These signals are affected by voltage and temperature as well, which are of course variable factors. I've tried to tweak the delays on a few occations, but that usually resulted in instant hangup, even though I was pretty careful. Given that any AM2 system performs this calibration, regardless of accessible BIOS features, and that manual adjustment didn't get me anywhere, I think they're pretty useless TBH. I mean, if it works for you, that's great, it's just that I have a different experience

Just to be clear, I'm mainly referring to the fine delays that are shown here:

http://www.diskusjon.no/index.php?ac...post&id=154879

[clon22]

oh great, thanks for the info.
I was wondering what was that option for too.

[Unoid]

Weaksauce

Why not unganged 2x128bit

we need 24mb/s throughput!


Amd should have focuesed on faster latency L2 and cut the L3. :P

[jarthel]

what the better between unganged and ganged?

would it somehow overcome the weakness (at least the reviews are united in saying it is slower than conroe) it presents against a stock q6600?

[Boschwanza]

To reply on this issue

http://www.hardwarecanucks.com/forum...e-preview.html

AMD Phenom X4 9600
Gigabyte MA790FX-DQ6 (BIOS updated to BETA F2i for stability at DDR2-1066)
Crucial Ballistix Tracer PC2-8500 @ 1066 5-5-5-18 2T 2.1V *Using "ganged" memory mode shows 128-bit bus in the BIOS and in AMD Overdrive, and gave better benchmark scores for gaming tests, so we used it*
ATi Radeon 3850 256MB (out of the box drivers)
Seagate Barracuda 7200.10 500GB SATA2
OCZ GameXStream 600W
Generic DVD-ROM

They tested in ganged mode which means 1*128 Bit. So normally they should get equal scores compared to a K8. But the canadians again experienced horrible memory results in ganged mode. I guess at this time ganged and unganged are just a placebo. K10 works with just one 64 bit IMC.

[Quintero]

They tested in ganged mode which means 1*128 Bit. So normally they should get equal scores compared to a K8. But the canadians again experienced horrible memory results in ganged mode. I guess at this time ganged and unganged are just a placebo. K10 works with just one 64 bit IMC.

You really should scroll up to post 5...