Printable View
More or less as Tony says...still testing...but what amazes me is that software can say any FSB...1,000 even if you wish to write it in the application but that certainly says nothing about whether any FSB will be actually bootable, useable or stable.
RGone...
Quote:
Originally Posted by Tony
ok, I'm not trying to piss anybody and if I did, I want to apologies.Quote:
Originally Posted by _33
But I do not see how all these rich features make the RD600 a top enthusiast board? Too much hype IMHO.
I just want to se some end users have the board.
I really hope it is what it is made up to be :)
Dont like the i680 :(
I wish I had read a post like that before I bougth my current motherboard :(Quote:
Originally Posted by Lestat
How lame is it to be limmited at 400 fsb....
And a DAMM fine point you maid there sir.
As long as all the boards we buy from e-tailers hit 500FSB with no issue's, I still say this will be the tweaker's board. I wanted to have one by now, but if DFI wants to delay to insure a working product with as few bugs & problems as possible, I can wait a few more weeks. Consider this, only the 965 chipset based motherboards seem to hit 500FSB consistently, with the trade off being performance vs 975 chipsets. The RD600 IMO has a lot going for it when you consider 500FSB with better performance than 965 chipset, low power consumption, and that tricked out bios. The 680i is a power hog, the chipset get extremely hot, its hit or miss hitting 500FSB (FSB holes), & extremely buggy (ps2 mouse & keyboard freezes & sata data corruption or loss). I don't recall these issue's being reported before the board was available.
I'm looking forward to owning and tinkering with this board. I've already gotten rid of my EVGA 680i to make room for it.:)
I think the issue will be be how consistently this board manages to hit those clocks. That's the variable we have seen on most other boards to date.
Most of the hype comes from the excitment of the users, some are ecstatic others have unrealistic expectations. Most of us including myself fill both.
Anything you see on the advertisiing front is the same we see on every motherboard, it's just this one has way more interest.
Bottom line of what I see is that people wanna buy a e6400 and overclock it to 4ghz, so the board must hit crazy high fsb's.
So now if a motherboard manufacturer can't provide that they're no good?
I'm afraid I tend to disagree, Intel were clever enough to lock those multipliers down for a reason, that's how the price range exists. Granted the difference we're talking about is an anal 100-200mhz or so? That's certainly true for the most part, less people run phase than those running water or air.
Best bang for buck for the more advanced mid range oc'ers is still the E6600, if you must get the extra 100mhz or so oc over the 2mb cache cores. That's a generalistaion not a hard and fast rule.
The rub is how stable that 511 is, if it can run unconditional at that fsb for 90% of users who have similar cooling and hardware config, that's will be a victory in itself.
The lanperty range has always been for tweakers and for the most part when DFI get it right, the boards ususally are more consistent clockers than any other. The rest is down to learnng curve..
Sometimes I wonder who really benefits from these threads as they seem to give all parties involved a hard time.
If the concern is really about spin, then we won't know the truth until we have the boards ourselves.
Best thing to do with a thread like this is to lock it so only those who have the board can post and make it reader only for the rest of us, might not keep people happy but that way we can keep our supreme intelligence to ourselves(me too)...
2-3 weeks from now the truth will be out anyway, I don't think anyone feels they'll drop a big bucket of donkey sh** on us all and then hide behind a lampost..
regards
Raja
Guys
All I see is people annoyed the board will only do 500fsb or so; all you have to do is buy clever with the cpu and your fsb issue is over.
Im testing badaxe2 at the moment, I have found a way of lowering memory latency by 4 to 7ns in everest, but doing so limits workable fsb on the board to 340 with a Conroe, maybe 345 with an Allendale. Now, couple this up with a 6800 running 12 multi and you are sat sweet at 4ghz or so and the ram is barely doing ddr900 with the performance of 1100+.
RD600 needs cpu's with higher multipliers, i think we all know that, but what you keep forgetting is you can pretty much dial the memory in at any speed you want. This alone makes this board quite a bit more flexible than the badaxe which is stuck with 1:1 or 3:4 etc etc for ram ratio's.
I firmly believe Allendales are built for 965 boards, Conroes are built for something thats that little bit faster like the 975 or the RD600 etc, but you are going to have to see 500fsb+ is pretty hard with a Conroe for any board.
I think it all boils down to how much hard cash you are willing to throw at your overclock on the CPU, if its sub $200 you really do need to look at 965 if you want to push 550FSB, If like me you need a little more horse power you need Conroe with that higher multiplier to get you there.
Has anyone done any testing with a 6700 retail yet?
If so, I would like to see some results as that is the chip I have.
I am also running ddr2-1000 modules with the D9's in them. D9GMK I think.
well said Sir Tony.. :)
OT: Merry Christmas :)
One thing that bothers me about this board is the dual 8x PCI-E lanes when in Crossfire. Didn't AMD eventually get dual 16x lanes working on their last crossfire board?Quote:
Originally Posted by Tony
Yep that is the classic tradeoff: Higher multipler gets you tighter timings and less latency all things equal. That being said it will be interesting to see how the E4300 (9x multi) and E4400 (10x multi) play with these boards.Quote:
Originally Posted by Tony
I am running the E4300 at 3.3 on this board without an issue. CPU is at 1.4250V currently. Kentsfield is in it now. ;)Quote:
Originally Posted by Phosphate
precisley my point all this spin has been created off the other boards namely the 965 variants, where people have posted some oc's of allendale up near 4ghz and now everyone wants buy cheap and get that performance and the mobo makers are bearing the brunt of the infinate fsb search.
E6600 is fine air and water upwards on 975. Even 980i's will work best with the x6800/E6700 as you remain under the strap changes for performance on air and water, if you can afford better cooling you should be looking better cpu too. I am not convinced that superpi times play any major part while I am playing 3d games, I get bemused when boards are asked to purely sell or be of merit based upon a single super pi run. The bigger market is definatley with the 3d gamers than the benchers and bingo laid out the percentage's for us all before. I should think that XS makes up a small percentage of the hits at anandtech reviews, the rest are gamers and other enthusiats thay have little concern for the ultimate 600fsb board, 511fsb is a god send to us all if it's stable for the majority.
There has been just too much spin of fsb in the industry, and I'm hoping that threads like these at least serve to bring some of us home for a reality check.
I know what you are saying about the high FSB for the allendale guys, but what I want to know is how much a said mobo can improve the total frequency potential of a CPU. I think Northbridge temps may play into that among other things (chipset timings etc). Haha, I'm no Expert.Quote:
Originally Posted by raju
Whenever I see a review of a Mobo and the guy puts it on the 6x multiplier and gets 550 FSB....I say to myself so what?:D That's only 3.3 Ghz
I think it's time we began to acknowledge that current engineering capabilities at this price point allow a max fsb and performance that has some give and take accross all the current chipsets.Quote:
Originally Posted by Phosphate
I have toyed with high frequency audio amps and psu's, and I can tell you this, that designers have to go to massive lengths to keep things stable at high mhz, oscillation is a killer at high mhz especially in a electrically noisy environment like a pc case with loads of psu rails all over the place. Trace inductance, ground planes, emi/rmi shielding the list goes on. Multi layer pcbs are a must, but that raises costs too, so we need a middle ground between both. An alternative to conventional glass fibre pcbs is available - Teflon pcb, but the cost would be astronomical.
People complain about hook n loop nb sinks but they have no idea why those loops are really used, study the pcb closly and in general you'll find that traces are run in the space that loops save over holes, Intel did it for a reason with the 975 and other chipsets have followed suit because it works better at the design level. The highest fsb strains are on the nb/ram/cpu connection, any savings there in terms of space and trace routing give the engineer a far easier time getting those crucial mhz that we all want.
R&D work is required behind the scene to take the mhz battle to the next level in terms of the upcoming nbridges that we don't yet know about, and thats gonna take time, question is how patient are we, and how much are we willing to pay if we are to get it right now?
Profit margins for ground breaking products are not always as high as we are lead to believe based upon immediate perception of individual component costs at pcb level.
regards
Raja
How high fsb have you guys reached when memory is running in sync mode?
The board / bios seem to have certain problems with sync mode.
Before finding the fix my board could do only 420FSB in sync mode.
The board I´m using is not a retail sample, just one of the release candidate revisions.
This board has a revision A13 NB, which is already fully functional revision so the sync mode problem should´t be NB related.
what fix was needed for sync mode clocking above 420fsb The Stilt?
Merry Xmas guys :)
Not much to see I know... but I spent the whole day just playing around in memtest86+ v1.65 with 2x512MB OCZ PC5400 Platinum XTC 3-3-3-12 to figure stuff out as there's so many bios voltages and settings to figure out. Once I got a comfortable and stable memtest86+ v1.65 config, I installed windows and proceeded to do some stock speed load tests to compare temperature readings with my E6600 air vs water results shown here.
System:
- E6600 QPGE B1 Step
- Tuniq Tower 120 with stock fan (High speed)
- DFI ICFX3200-T2R/G 12/23 beta bios
- 512MB HIS X1800XT + VF900CU
- 2x512MB OCZ PC5400 Platinum XTC 3-3-3-12
- 74GB WD Raptor 8MB
- Pioneer DVD Burner
- 1.44MB FDD
- 1KW PCP&C SLI psu
- WinXP Pro SP2
- SB600 Catalyst 6.11 Chipset driver build 37616
System pics in sig linky ;)
Under Load Digitech QM1538 Digital Multimeter measured psu rails as follows:
Rail = Smartguardian = DMM real
+12v rail = 12.09v = 12.26v
+5v rail = 4.89v = 5.10v
+3.3v rail = 3.28v = 3.368v
DFI ICFX3200-T2R/G 12/23 beta bios settings used:
(additions since 12/18 bios in bold)
http://fileshosts.com/intel/DFI/ICFX..._validated.pngQuote:
Memory Clock Frequency: Sync
Current Memory Clock: 266
Control CPU Frequency: Enabled
x CPU Frequency (Mhz): 267
NB FSB Strapping: AUTO
PCI-E Overclock: 100
Software Reset Clockgen: Disabled
Memory Overclocking: Disabled
CPU Vid Control: 1.25v
CPU VID Special ADD:
DDR2 Ram 1.8 Voltage: 2.15v
CPU VTT 1.2V Voltage: 1.31
NB Core 1.2V Voltage: 1.36v
NB PLL 1.8V Voltage: 1.83v
NB PLL 1.2V Voltage: 1.38v
NB PCI-E 1.2V Voltage: 1.34v
NB Core 1.2V convert from: 1.89v
SB Core 1.2V Voltage: 1.34v
Clockgen Voltage: 2.92v
GTL REF Voltage control: Enabled
CPU Core 1/2 GTL REF Volt: 48
North Bridge GTL REF Volt: 48
CPU PLL Setting Overide: 128
PLL Override loop count: Enabled
Conroe P-State Multiplier: x9
C1E Function: Disabled
CS Output Timing: Delay 1/4 clock
CK Output Timing: AUTO
Dual channel Split Mode: AUTO
ADDR/CMD Driving Pre-AMP: Enabled
ADDR/CMD Driving Strength: 15
Clocking Driving Pre-AMP: Always
Clocking Driving Strength: 5
Address/Command Timing: 1T
Memory Timing Select: Manual
CAS Timing (TCAS): 3
Read RAS-CAS Delay (TRCDR): 3
Write RAS-CAS Delay (TRCDW): 3
RAS Precharge Time (TRP): 3
Cycle Time (TRAS): 8
Row Cycle Time (TRC): 18
Refresh to Active (TRFC): 35
Bank to Bank CMD (TRRD): 3
Write Recovery (TWR): 5
Read-Write Command (TRTW): 3
Write-Write CS (TWTW_CS): 2
Write-Read Command (TWTR): 3
Write-Read CS (TWTR_CS): 2
Read-Read CS (TRTR_CS): 2
Read Data Transfers: 5
Memory Auto-Calibration
Read Data Transfer: Enabled
Read Strobe Delay: Disabled
Read Strobe Arrival: Enabled
CPU Idle:
http://fileshosts.com/intel/DFI/ICFX.../cpuz_idle.png
CPU Load:
http://fileshosts.com/intel/DFI/ICFX...ime95_load.png
CPU Post Load Idle:
http://fileshosts.com/intel/DFI/ICFX...tload_idle.png
The DFI ICFX3200-T2R/G seems to have a higher idle temp than my experiences on Asus P5W DH Deluxe, Asus P5B Deluxe, Asus 680i Striker Extreme and Intel D975XBX Badaxe. All other boards were also tested with Intel Speedstep and C1E disabled like DFI ICFX3200-T2R/G. But load temps are on par for air cooling. Rise above ambient delta temps seem to agree with my E6600 air vs water cooling tests with Intel D975XBX http://i4memory.com/showthread.php?t=3218 although at a lower vcore than on Intel D975XBX Badaxe :)Quote:
E6600 @2400Mhz at 1.23v
DFI ICFX3200-T2R/G
Tuniq Tower 120 with stock fan (High speed)
Room temp: ~28.2-30.8C
Cooling used: Tuniq Tower 120 with stock fan (High speed)
Bios idle: 43-48C
Windows idle: 37-40C
Windows load: 50-51C with fluctuations in coretemp down to 48C
Rise above ambient delta = 19.2-22.8C
Rise above windows idle delta = 10-14C
Guys just be aware that there's ALOT of settings to test and it takes time... I've only had the board up and running for less than 24hrs so far and haven't tested alot of stuff stillQuote:
E6600 @2400Mhz at 1.259v
Intel D975XBX Badaxe
Scythe Infinity + 2x Thermaltake 120x25mm 81cfm push-pull
Room temp: ~27-29C
Cooling used: Scythe Infinity + 2x Thermaltake 120x25mm 81cfm push-pull
Bios idle: 40-43C
Windows idle: 35-36C
Windows load: 50-51C
Rise above ambient delta = 22-24C
Rise above windows idle delta = 15-16C
Just a simple bios trick that had to be done on the reference RS600/RD600 boards too.Quote:
Originally Posted by flytek
I want to hear other experiences before telling more.
Just in case the problem is related to this specific board.
The board works now atleast up to 480FSB in sync mode.
what bios you on ?Quote:
Originally Posted by The Stilt
12/18.Quote:
Originally Posted by eva2000
Got pm btw ;)
I could not get over 416 in sync mode until this last BIOS (12/23) update. I am at 476 stable now, almost hit 500 but stability was an issue. I played with the new BIOS settings for a few hours last night and realized since the 12/07 BIOS that about 40% of my original settings need fine tuning again. :D I was unable to run higher than 7x506 with the E6600 on this BIOS (12/23), was easily hitting 7x511 stable before. I also dropped from 9x432 down to 9x426 but I was able to drop VCore from 1.5375V down to 1.5000V in the process (yes, my E6600 is not a good sample). However, a few benchmarks have improved a couple of percent so far with BF2 coming up a couple of frames but HL2 dropping a couple. Even so, this board will not be a benchmark record holder but I still like it. :clap:Quote:
Originally Posted by The Stilt
EVA-
My Northbridge temps peak around 58C under load with the stock heatsink after removing the pad and using a little less paste than DFI placed on it. I was up around 61C completely stock, throwing some addtional air across it dropped my temps down to 52C now. RGone found that adding a small heatsink to the clockgen will also help. I plan on trying it today along with a modded SwiftTech cooler for the Northbridge.
I also found that depending upon whether I want to run 1T between 800~900 or run 1200 at 4-4-4 timings that the board is very sensitive still to the memory module I use. I will PM you later today with my working list to date after some additional testing.
I'm quite skeptical about wether DFI will leave this board all passive cooling. I mean even the LP NF4 boards have active cooling. I would really look forward to an active cooling on the chipset, and maybe some extra heat sinks where it counts most. I believe it makes sense for this tweaker board to have that sort of help instead of giving that extra hassle to the buyer :confused: , specially considering this isn't a value product.
I still own a LP NF4 UT board and am looking seriously into switching C2D + ICFX3200 in the near future, as it is starting to make some sense. Also I am wondering if there will be any derivatives of this board? An Expert or Venus model?
Anyhow, this is getting interesting none the less! :p: :clap:
Sorry for my noobness but which IC is the Clockgen IC?Quote:
RGone found that adding a small heatsink to the clockgen will also help. I plan on trying it today along with a modded SwiftTech cooler for the Northbridge.
Quote:
Originally Posted by raju
I agree completely. People need to understand what you are giving up to get those few extra FSB. While it is nice that I have a couple of ASUS boards that do 520FSB with the E6300, I would much rather have an E6600/6700 on a good 975X board at 415~420FSB. Our discussions with the motherboard suppliers center around stability and peak performance at given FSB points and not on producing a board with the highest FSB in order to grab headlines. It just seems as if everyone has gotten into the FSB race and decided to ignore actual performance or stability for the sake of a good SuperPI or CPU-Z screenshot. Unfortunately, unless you show this capability people will say the board sucks.
The people complaining about the FSB limit on this board should realize at this time the board is more stable at its upper limits in actual use than any other non-Intel chipset (better than most of those) board available. While it does not provide the best performance in benchmarks, it does give the user great flexibility with their component choices.
In the end, not a single one of us would be able to tell the difference between this boards performance or the 680i/650i/965/975X without a benchmark in day to day operation. What matters is stability, features, support, warranty, and price in my opinion. I would just assume to throw out most of the bencmarks and look at overall platform stability at given FSB/Memory settings. For the benchmarks left, they should be compared on a clock by clock basis between chipsets. A wider audience might start to fully realize that you need that additional 35~50FSB on a 965 board above 400FSB to make up for performance losses due to memory strap changes as an example.
I would be a lot happier to have a stable P965 system at 390FSB than 450FSB any day due to the memory/mch timing improvements. When the first P965 boards were introduced, the suppliers were working towards that goal as they knew the best performance of the chipset at a reasonable cost was to get into the 385FSB range in a stable manner, people started complaining about not having high FSBs once they realized the overclocking headroom in the Core 2 Duos, so now we get 450FSB on average but did performance really improve that much? The new target for any "decent" P965 board is 500FSB and with it we get higher board costs, less stability overall, and lower performance at stock or near stock speeds just to name a few issues.
The problem is we expect our components to achieve at least a 50% improvement in speed without question. It is now expected this capability will be provided to us at no cost. Do we expect this from the automobile manufacturers? We pay extra for that level of performance. While I believe asking $1000 for a CPU or $400 for a desktop motherboard is way overboard the majority of time, I think it is wrong to expect, now demand, that same level of performance from a $150 CPU or $65 motherboard. Rant over... :D