|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
41.183.0.21
In Reply to: RE: cMP - the open source high-end Memory Player posted by cics on December 30, 2007 at 05:42:01
After a lengthy evaluation of nearly 2 weeks, I've gone back to my 45nm setup (E7200 + Kingston HyperX DDR2 + GA-G31M-S2L) as per recommendation.
Seger summed it best here. My experience is the same (i.e. dull veiled sound) even with lower RAM voltages of 1.300V and Vreff/Vaddress/Vdata=0.550/0.540/0.540.
Temperatures are very low with 7x24 playing yielding temps of only 34°C doing 192k@145db. Although power consumption would be lower it seem this GB mobo (GA-H55M-UD2H) is a poor choice for audio (it's a gamers board with fantastic BIOS control). There's too much crammed into it and there's no "clean" PCI slot. Perhaps LGA775's *-S2H mobo is a better choice.
Follow Ups:
So I did a comparison tonight with both machines using power supplies with the fans removed, and the Core i3 using a separate ATX PSU for P4.
The Core i3 machine is BIOS optimized, and you can't do much with the Atom, but it's designed for low power.
The Core i3 wins - bigger soundstage, more air, a little more musical.
I've tried a couple of set ups too. Reference can be made to the results below for choosing MB and CPU.
Core i3 530(GA-H55M-UD2H) > E7500(GA-G31M-S2C) > GA-D510UD > an Asus Netbook
with Atom D450.
The same PS was used on the first three combos.
I believe the power supply circuit and BIOS flexibility on the MBs have significant impact on the SQ.
After switching back to 45nm technology, I was surprised with the sterile sound and fortunately found the culprit! These new test changes (especially gpedit) were the cause. Also, an important BIOS setting not available in the H55M-UD2H mobo, PCI Latency, had to be addressed.
I reinstalled my 32nm setup with following changes:
- Upgraded to latest BIOS - F9d
- Used recommended settings as per site
- Used PCI Latency Tool 3.1 to set soundcard latency to 128. This adds a new service and requires BIOS devices under "Integrated peripherals" to remain enabled except for IDE, Serial Port and Sata ports 0-3. Remaining unused devices are disabled in Windows. Disabling other "integrated peripheral" devices in BIOS causes PCI Latency Tool to not recognise PCI devices.
- Vcc = 0.72500 (900MHz), Vgcore = 0.650 (400MHz)
- Vdram = 1.400 (600MHz), Vref = 0.600, Vdata = 0.560, Vaddress = 0.560
- Dram Timing = CL5 & Command Rate = 1
Sound quality not only meets previous best 45nm setup but surpasses it beautifully! I'm going to leave this setup for a while and see if there's further improvement. There's also another mobo on its way which hopefully removes the PCI Latency Tool.
Zalman cooler (with fan removed) does a great job. Temps are steady at 36°C (previous 34°C was achieved using Vcc of 0.70000V or one step lower).
Is this something special?
Please advise.
THANKS,
Rick McInnis
.
pci latency tool makes the sound too analytical! somehow i did not like it..
its not music that i want..
ASIO latency setting is very important - set this to lowest possible. For PCI latency at 128, ASIO latency must less than 64. What is your ASIO latency?
For ASIO@64, test PCI latency at 136 or 144. For ASIO@96, test PCI latency at 200 or 208. Always keep PCI latency in multiples of 8.
I'm glad you've been able to find a solution.
Do you think a future Bios update will address this PCI devices problem?
Are you also implying that "Autoplay changes" through gpedit should not be performed in an old system?
Thanks
Can you give perhaps an overview on what new stuff is required besides the processor, ram and case (items plus general cost...approximate)? Also I know you say it is better but for the record...is it a lot better?
Trying to assess whether I should embark on this new path.
Cost is ~$300 for Core i3-530, H55M-UD2H mobo, Kingston DDR3-1375 1GB RAM and a 1156 adapter for mounting fanless heatsink.
I want to compare this mobo to an alternate cheaper option. This should remove the need for PCI Latency Tool and its simpler design (i.e. less mobo features) may improve SQ.
Is it a lot better? Lets put it this way, I'm never going back to 45nm technology! Whilst there's no new detail, the big improvement across the board is how well low-level detail is fleshed out. There's harmonic richness that's very pure and beautiful. Also a bit more air and lower noise floor. Sounds emerge from the deepest of blacks I've ever heard. Damn this sound is addictive.
I'm testing different cPlay buffer options as I expect Tiny option to offer best result with Core i3+192k SRC@145db - 3 cache layers where Tiny option reduces L2 cache footprint allowing for HT to work better.
This is great. I have been waiting for your final words to build a 'real' music server to replace my current Aspire One netbook + Devilsound dac. I must say that the netbook has been giving me immense musical pleasure for the past year.
Ok thanks I'll wait for your update on the cheaper mobo alternative.
.
I just added back my Juli@ with unbalanced analog outs to my UD2H system and I'm liking it. My experience says the stock Juli@ digital outs aren't that great with this board, but analog sounds quite good. Or maybe the stock breakout cables are the main problem with digital out. For a while I wanted to avoid a DAC to minimize jitter, and now with a good low voltage board, I'm back to that approach. Nice PRAT, detail, musicality.
I can't compare with the S2L board as I never had one.
Now I tried Juli@ analog in the Atom box and there's a nice improvement over the UD2H, particularly with symbols and acoustic guitar - HF sounds are much more evident in the mix.
I'm starting to prefer the least possible jitter-induced distortion compared to that produced by op-amps, etc. if I have to choose between an external DAC or lower quality sound card. I have a good silver coax cable and a good USB cable and the V-DAC/Pinkie is good, so at least to my ear (brain), there seems to be an inherent advantage to producing analog signals in a very low-power computer, one that's closer to the power needs of a DAC.
Still experimenting,
Dave
I'm very impressed with my Core i3 after fixing a few important things. Have you done a similar setup for it?
It would be interesting to compare Atom D510 vs Core i3. In both case power consumption is very low. Intel plans to release a 32nm version of the Atom.
I downloaded the F9 BIOS but haven't installed it yet. What's new with F9 that you can see? I'm on F8 still.
After cics reported dull veiled sound with his Core i3 combo, I performed some listen test (again) bewteen my "GA-H55M-UD2H+Core i3-530 32nm+Kingston Value Ram DDR3-1333 1GB" and "GA-G31M-S2C+E7500+ADATA Vitesta Extreme Edition1GB". I have learned and benifited a lot from cics' contribution as well as this forum, therefore I feel obliged to share my findings and experience here.
Test Conditions
Bypass capactors were added to almost each and every Cap. underneath both MBs(this offerred good SQ improvement). I am using a DIY USB DAC (battery drived) and there is no issue with PCI interface. The same HDD and PS (a 90W pico ATX ps and a pure battery + regulator power supply) were used for the listening tests. The rest of the system is all DIY. Bios setting on the MBs are similar to those used cics. The Windows XP Pro SP2 is fully optimized as per the instruction on this Homepage.
Results
After listening to the two MB combos for a few times, a couple of friends and myself unequivocally considered that the Core i3 530 combo sounded better all round. With the Core i3, the imaging is better and the sound stage is deeper and more three dimensional. The piano and vocal sounded more natural and real, highs are more detailed and lows are more clear and authoritative. There are more ambience Power consumtion appeared lower by a margin as reflected by the temperature of the power supplies.
I don't know what is the cause of the difference in SQ between my combo and that of cics, perhap I didn't use any PCI slot, but according I what I've heard in the tests, the Core i3 combo does sound better than the G31+E7500. The diffence is more than marginal, I feel like listen to two systems, one with good sound quality and the other very good.
One thing I experienced with the Core i3 combo is that it has better rejection to poor power supply quality, perhaps due its sophisticated onboard swithcihng supply circuit and a lots of Oscon capacitors on the MB. When I first fire it up, no bypass capacitors were added to the MB but it already out did the G31+E7500 combo. However, when battery power was suppled to the two MB combos, there is more improvement with the SQ of the G31+7500 than the Core i3, although the latter still won.
Hope someone can do more tests on the Core i3 and other MB.
When I say dull / veiled SQ, take this into context of my 45nm setup. Another way of describing this sound is analytical and uninvolving. Overall there's the same amount of detail but it lacks that finer resolution and impact.
The key issue is PCI slots which are wired differently (more complex) and perhaps the *-S2H option does not suffer from this problem. Your experience using USB will be different.
Anyway, I'm planning to try another mobo which may offer improvement. Core i3 + DDR3 is a big step forward especially ito power consumption (7x24 playing temps were 34°C vs 45nm setup at 45°C). The mobo choice remains a challenge for now...
I had J&W Boards in the past for oc and was really stunned about the build quality, especially their phase design and current stability when overclocking. The mini itx board of this brand called mini x was the best build itx board i´ve seen, unfortunately it is AMD only. But... they will release a Intel G45 mini itx board soon, this could be a killer mobo. Like DFI they offer lots of bios options for fine tuning in bios.
It looks pretty good. Atom 330 Minix is very well built and I definitely like the P4 connector - it's a must as most vendors drop this. Current Intel boards offer Atom 230 & 330 using the horrible 945 chipset.
I'll definitely try their Atom D510 option which is likely to be built (not yet available).
Hi cics
Afaik it´s a small company, the board designers were working for Abit before wch is legendary for its quality. I think that we may could take some influence to some bios options wich we need for cMP². I could ask the german guy who´s in charge of the distribution over here, he´s in tight contact with the asians and possibly can tell if it´s possible or not.
Would be in his interest beeing the exclusive seller over here for J&W, with the Klang & Ton cMP² series in mind this could create a win win situation.
I'd love to see 2 options: H55 (supporting core i3) and NM10 (for Intel's new Atom processors). In both cases all unused devices including onboard soundcard must be removed, we need flexibility on powering options (e.g. jumper to supply PCI slot with external 5V supply, etc.) and only 1 DRAM slot. I'm sure we can list more 'must haves'.
I'd pay 2-3 times the price of a OC capable mobo as long as quality and minimalism is maintained.
How about the existing 1156 itx boards? They (afaik) only come with one pcie slot, however, so the "mainstream" cmp cards can no longer be used.
"We should no more let numbers define audio quality than we would let chemical analysis be the arbiter of fine wines." N.P.
I hope you don't give up and have a chance to test on GA-H55M-S2H.
I believe lga775 has it.
Good listening!
is the problem only in dull veiled sound? or anything else?
if there is please state it.. so others will know.
See here for system - soundcard now used is Lynx AES16.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: