Phenom 9500 w/ MSI K9A2 Platinum

[KTE]

Mathos: yeah, your read data is very slow for MEM and IMC. Everything else is good. I've never ran them latencies so I'm not sure why.

As I tested quite long ago, after Macci showed the main bits, we found memory/IMC performance jumps with core speed (as well as cache performances obviously).

would going form 1800, to 2g make much of a differnce?

Depends in what exactly you want the difference to be seen in. Overall, yes, there is a slight difference, but this depends on the coding of the app. you're interested in seeing the difference.

HT speeds doesn't make too mucha difference unless you have powerful cards, whereas it does but so does PCIe speed. 110-115MHz should be easy on this chipset.

Morais: I have no idea yet.

[KTE]

You guys deserve an update. My trusty Phenom is dead: http://www.xtremesystems.org/forums/...0&postcount=64

Beware of IMC/MEM volts! I did not see any problems until I started running a larger delta between IMC and MEM volts.

i.e. 2.196V MEM and sub 1.25V IMC (> +/- 0.954V).

Worst of all, my gorgeous 770 and 780G princesses are both lying their cherry virgin, dressed in multi-colored copper, latches wide open, naked to the bare flesh, without a sign of penetration by the [p]rick Phenom. He, the royal stud, is not complying.
All my P4/K8/K10h/C2D/C2Q comparitive tests, roughly 41 individual tests which were 70-98% complete are now meaningless once again.

I will not test another Phenom, the replacement will most likely be sold there and then [1]. The hassle of 3 dead Phenoms already isn't worth it and having to start learning the intricacies and oddness of yet another one is a time consuming pain... I can bear but have no time for. I have already spent more time on this than I had available, and I have a life to live now. You guys will have to carry on Phenom+MSI RD790 yourselves, sorry.

[1] Now let's hope I stick by that statement.

The one I will test in Abit 770 and a 780G, and keep for personal use replacing my BE, is:

Athlon X2 4850e ADH4850IAA5DO 2.5GHz 65nm G2 45W TDP, CnQ @ 1.0GHz 1.1V 18.1W TDP - > $89

The other two low TDP 65nm G2 dual cores to release will be:
Athlon X2 4450e 2.3GHz ADH4450IAA5DO (replacing BE-2350)
Athlon X2 4050e 2.1GHz ADH4050IAA5DO (replacing BE-2300)
[Athlon X2 BE-2400 (2.3 GHz) - > $78 (now $104), Athlon X2 BE-2350 (2.1 GHz) - > $68 (now $91)]

More upcoming releases late March and early Q2 (April):

Athlon 64 X2 5400+ 2.8GHz 65nm Brisbane G2 - 65W
Athlon 64 X2 5600+ 2.9GHz 65nm Brisbane G2 - 65W and 89 W
Athlon 64 X2 5800+ 3.0GHz 65nm Brisbane G2 - 89W

[LIKMARK]

Thats too bad, I'm hoping you wont stick by your statement

[madfaze]

sorry to hear that KTE...


anyways how can you measure TDP?are there any software that i could possibly use?

[Morais]

P=v.i

[madfaze]

P=v.i

how is that?

are there any software that i could use?

[Morais]

I dont think so... I think the best way measuring is with a multimeter, then do the maths...

[KTE]

TDP using above formula if you know the exact current your CPU is pulling (encoded within MSR) or you just trust the manufacturer.

TDP is thermal heat dissipated rather than power draw. Power draw you can very easily measure within a good approximation using a few electrical tools.

The Phenom 9600 and below need 76A. So look at mine, 1.232V @ 76A giving 93.6W TDP.

Top rated bin: 1.25V x 76A = 95W TDP

But the B2 Phenom 9900 required 110A for the clocks. That's where the big difference in their TDP, power draw and temps. comes from which users have conjectured over many times.

That's why if you run your < 9600 BE at 2.6GHz stable, it won't reach the same power/temps as the 9900.

As for Phenom being hot, not so. I just checked another myth...

Stock cooler is not capable of dissipating heat which it draws away from the IHS, so heat becomes trap within the heatsink and aggregates causing a heatbed to form and so the core heats up more and more.

I tested Phenom 9600 BE at stock without any cooler or fan, 24C ambient.
I have already established mine switches off at 72C BIOS repeatedly.
I then had it loaded at POST (one core 100%).
Kept thermocouple at the top and sides.
After 25 seconds, thermocouple read 55C IHS and 59C at the side nearest die (hottest bit).
I then shut it off and to test if the temperature is more or lower than that read. I quickly took it out into my hand when thermocouple read 53C surface.
Measured at underside pins and it read 48C.

Yep, it was definitely around 50C, since it was very warm and low-hot to the hand holding it but not burning. I held it minimum 14 seconds and it had dropped to 50C IHS. If you are a medic, medically educated or an electrical engineering individual then you would know that by touching 63C metal for 1 second, you would be burned, 58C for 4 seconds and 55C for 10 seconds. 48C will not burn your skin until 10 minutes holding it.
[i.e. Ref. BS 4086:1966 Recommendations for - Maximum surface temperatures of heated domestic equipment, PD 6504:1983 [BSI published document] Medical information on human reaction to skin contact with hot surfaces, BS EN 563: 1994 Safety of machinery - Temperatures of touchable surfaces - Ergonomics data to establish temperature limit values for hot surfaces]

This is obviously inline, showing that the temperature even without a heatsink and fan took 25 seconds to rise to 60C, one core full load. That would show that it won't be 40C idle at stock with a heatsink and fan unless one of the components is underpowered for 95W TDP, or can just about remove 95W TDP and keep the CPU below HTC activation.

I tried the same but with a Q6600 at 3.6G 1.344V just to recheck. It jumped to 75C IHS under 11 seconds and it shut off (THERMTRIP#) before I could switch it off manually, under 16 seconds (105C). It was completely untouchable at 63C IHS reading, would burn me straight away. That was just to recheck the readings were OK.

[sam2008]

Sorry to hear that KTE. Too bad. Any peticular reason that cause it to die? Too much heat built up or just too high core voltage? Maybe you should use a better HSF(water cooled) to avoid overheating. So far, you are the only one that I know of has such a bad luck though.

You guys deserve an update. My trusty Phenom is dead: http://www.xtremesystems.org/forums/...0&postcount=64

[Shadowlite]

Hi guys, I need help, I do not intend to o/c my system, thus i just want to know what settings to set for a pc8500 (1033mhz) kingston hyperX ram. This is because Kingston's website only state 5-5-5-15, but the manual settings at the bios has other options for me to set.

I think that's the cause of the problem in which the board detect the 2 ram as single channel, and not dual channel, even though I installed it correct according to the manual.

Thanks for the help in advance.

My rig config is :
Phenom 9500
MSI K9A2 platinum
PowerColor HD3870 OC
Coolermaster Realpower Pro 750W

[Mathos]

TDP using above formula if you know the exact current your CPU is pulling (encoded within MSR) or you just trust the manufacturer.

TDP is thermal heat dissipated rather than power draw. Power draw you can very easily measure within a good approximation using a few electrical tools.

The Phenom 9600 and below need 76A. So look at mine, 1.232V @ 76A giving 93.6W TDP.

Top rated bin: 1.25V x 76A = 95W TDP

As for Phenom being hot, not so. I just checked another myth...

Stock cooler is not capable of dissipating heat which it draws away from the IHS, so heat becomes trap within the heatsink and aggregates causing a heatbed to form and so the core heats up more and more.

I tested Phenom 9600 BE at stock without any cooler or fan, 24C ambient.
I have already established mine switches off at 72C BIOS repeatedly.
I then had it loaded at POST (one core 100%).
Kept thermocouple at the top and sides.
After 25 seconds, thermocouple read 55C IHS and 59C at the side nearest die (hottest bit).
I then shut it off and to test if the temperature is more or lower than that read. I quickly took it out into my hand when thermocouple read 53C surface.
Measured at underside pins and it read 48C.

Hmmm, that means with my Phenom running at 1.240v according to everest and everything else that can pick up voltage, that I'm running at 94.24W while I run at 2.6 by 2.4? Going by that since the 2.4nb/IMC is at stock volts. Hehe, kinda liked my stable or no post luck with my Nb/imc on this one.

Yeah I kinda figured that with the stock cooler, thats why I didn't bother with it, and just put it on the old x2 4200+ 65w Windsor. Though it could explain why I've been able to OC at much lower voltages than a lot of people, probably due in part to using the ZT Nirvana for the cpu cooler.

[KTE]

sam2008: my last 3 Phenoms had same problem. It's caused by a bad IMC, suddenly it bunks out and won't boot and load the memory. Achim's 9500 has the same problem after a while, he even had the same RAM as mine with similar volts/timings, so did a few others here. The problem as I suspect very much from testing since November, is caused by keeping a large delta in between RAM and IMC volts mainly.

No pushing or overheating, volts were low. I only kept low volts for over a month now, sub 1.3V CPU sub 2.2V MEM and sub 1.1V IMC. It was stock volts for a day. It would never happen if I did not clear CMOS, I know this from testing. With me, it happens only when you clear CMOS, each time, repeatedly and I always dreaded doing so.

[KTE]

Hi guys, I need help, I do not intend to o/c my system, thus i just want to know what settings to set for a pc8500 (1033mhz) kingston hyperX ram. This is because Kingston's website only state 5-5-5-15, but the manual settings at the bios has other options for me to set.

I think that's the cause of the problem in which the board detect the 2 ram as single channel, and not dual channel, even though I installed it correct according to the manual.

Welcome

What exact RAM do you have, can you check in CPU-Z and post a screenshot of the SPD tab back?
Are they 512MB, 1GB, 2GB or 4GB modules?

First set both modules in the slots next to each other (green+orange). That will be dual channel (manual is wrong).

If it's 1GB or less modules then set the RAM tRFC to 105ns.

Which BIOS are you using? Don't try 1066 RAM with any BIOS before P0E.

Hmmm, that means with my Phenom running at 1.240v according to everest and everything else that can pick up voltage, that I'm running at 94.24W while I run at 2.6 by 2.4?

When you oc speed/volts, speed affects TDP but voltage affects it by the square. So if using a constant current, you would have to use the formula I posted before to work out your new TDP: Pn = Pt.(vn/vt).(Un²/Ut²)

dr_drache: I checked your logs through. You have the famous bugcheck:

Bug Check 0x101: CLOCK_WATCHDOG_TIMEOUT

The CLOCK_WATCHDOG_TIMEOUT bug check has a value of 0x00000101. This indicates that an expected clock interrupt on a secondary processor, in a multi-processor system, was not received within the allocated interval.

It means your processor has become unresponsive i.e. locked. It's caused either by instability (lack of voltage/heat buildup/bad PSU rails/volts), software incompatibility or software-driver-OS problem. It could be anything but it is looking most like a Vista x64 issue since it references the 64b kernel and kernel modules in the stack just before the problem.
I'm going to get some more symbols and debug it a little more later on. Need to update some files.

[sam2008]

KTE: Why clearing the CMOS has something to do with it? Is it because this peticular MSI motherboard? Maybe switch to another motherboard won't cause the death of your phenom.

Come on KTE, we need you to continue staying with us . Don't give up on Phenom. If phenom is that easy to OC like Q6600, it is not fun at all isn't it?

sam2008: my last 3 Phenoms had same problem. It's caused by a bad IMC, suddenly it bunks out and won't boot and load the memory. Achim's 9500 has the same problem after a while, he even had the same RAM as mine with similar volts/timings, so did a few others here. The problem as I suspect very much from testing since November, is caused by keeping a large delta in between RAM and IMC volts mainly.

No pushing or overheating, volts were low. I only kept low volts for over a month now, sub 1.3V CPU sub 2.2V MEM and sub 1.1V IMC. It was stock volts for a day. It would never happen if I did not clear CMOS, I know this from testing. With me, it happens only when you clear CMOS, each time, repeatedly and I always dreaded doing so.

[Mathos]

Welcome

What exact RAM do you have, can you check in CPU-Z and post a screenshot of the SPD tab back?
Are they 512MB, 1GB, 2GB or 4GB modules?

First set both modules in the slots next to each other (green+orange). That will be dual channel (manual is wrong).

If it's 1GB or less modules then set the RAM tRFC to 105ns.

Which BIOS are you using? Don't try 1066 RAM with any BIOS before P0E.

When you oc speed/volts, speed affects TDP but voltage affects it by the square. So if using a constant current, you would have to use the formula I posted before to work out your new TDP: Pn = Pt.(vn/vt).(Un²/Ut²)

dr_drache: I checked your logs through. You have the famous bugcheck:

Bug Check 0x101: CLOCK_WATCHDOG_TIMEOUT

The CLOCK_WATCHDOG_TIMEOUT bug check has a value of 0x00000101. This indicates that an expected clock interrupt on a secondary processor, in a multi-processor system, was not received within the allocated interval.

It means your processor has become unresponsive i.e. locked. It's caused either by instability (lack of voltage/heat buildup/bad PSU rails/volts), software incompatibility or software-driver-OS problem. It could be anything but it is looking most like a Vista x64 issue since it references the 64b kernel and kernel modules in the stack just before the problem.
I'm going to get some more symbols and debug it a little more later on. Need to update some files.

That bug right there sounds a lot like Errata 254, possibly 298, which may explain why MS applied a software TLB fix in Vista SP1. So if the voltage delta between memory and IMC also causes an issue, then perhaps 1.8v memory was the way to go?

In other news though, I got my new Toxic Edition 3870 installed now, with 8.3 drivers. Not have any of the issues I had with the other card, even while having the CPU OCed.

Hmmm finding the actual values for current, could be difficult since it can't be measured properly without making the Multimeter a part of the circuit. But it does make sense that the actual current in the die usage could be 67amps. After all the hard input to the cpu out of the PSU is 12v, and draws about 6.7 amps off the 12v rails. Considering a CPU staging area is a massive parallel circuit, which if set up right divides voltage, and after the voltage is divided, you could put a current amplifier circuit off of each branch. When I was doing electronic classes in such in college it wasn't difficult to set up an amp circuit that could take 200mA in and output 2Amps, granted in those circumstances the voltage wasn't lowered so soon after the resistors and transistors would start to let the magic operating smoke out.

[dr_drache]

@KTE

thanks, if there is some "expermental" way to fix this crap :P, let me know, i'll spend what time i have to test stuff if you think of a way to attempt to "fix"

im any case it looks like i'm about to buy some ram, don't feel right running one stick with 1/2 heatspreader (and only one heatpipe)
soo, the 1.33 bios is good for 1066?

[Shadowlite]

thanks KTE,

I am using 2* 1GB Modules HyperX PC8500 from kingston. I installed the ram into the first 2 slots, orange and green nearest to the cpu :P ( knew the color coding was wrong from a review ). I am using the official 1.3 bios. under CPU-Z, my timings are:
Memory Type DDR2
Memory Size 2048 MBytes
Memory Frequency 533.4 MHz (3:8)
CAS# Latency (tCL) 5.0 clocks
RAS# to CAS# (tRCD) 5 clocks
RAS# Precharge (tRP) 5 clocks
Cycle Time (tRAS) 15 clocks
Bank Cycle Time (tRC) 14 clocks
Command Rate (CR) 2T

[Daveburt714]

I will not test another Phenom, the replacement will most likely be sold there and then [1]. The hassle of 3 dead Phenoms already isn't worth it and having to start learning the intricacies and oddness of yet another one is a time consuming pain... I can bear but have no time for. I have already spent more time on this than I had available, and I have a life to live now. You guys will have to carry on Phenom+MSI RD790 yourselves, sorry.

C'mon KTE... 9850 BE's are only a couple weeks away.... Is she a go'er know what I mean, nudge, nudge, .... Uuaaa....

http://www.xtremesystems.org/forums/...d.php?t=180276

Just what I need, another Phenom, but I'll probably get one it the price is reasonable...

[Morais]

KTE, AMD started to use a new nomenclature for dissipating power in phenom CPU's...its called ACP(Average CPU Power).Real TDP should be around 120W.

[KTE]

News from MSI BIOS Engineers yesterday... they're working h-a-r-d to add NB Multi options right now.
I'll have a BIOS soon they say, but I've returned my Phenom already.

See how things can change Jon

KTE: Why clearing the CMOS has something to do with it? Is it because this peticular MSI motherboard? Maybe switch to another motherboard won't cause the death of your phenom.

Tried 3 different boards, same problem. The problem is Phenom IMC failing to initialize (as in signs of being dead).

Come on KTE, we need you to continue staying with us . Don't give up on Phenom. If phenom is that easy to OC like Q6600, it is not fun at all isn't it?

Kinda true. I don't enjoy working with K8/P4/C2/Penryn, it bores me. There's nothing to do. But I did enjoy playing with Phenom, monstrous options and tweakability there, day by day it only got better and better for a true enthusiast not after records, and searching to tweak, learn and discover. If only onliners would use all the available tools, most of them are only skimming on it's oc options.

On another note, I also hate it's instability, randomness and sudden deaths, I have to say. I had just started finding the sweet spots for this CPU and its running as I wanted and suddenly, it just blanks out.

Also Phenom does better in Vista than an Intel part does from moving, the deficit for Intel moving from XP>Vista is larger than for AMD/ATi due to driver support and optimizations mainly, especially in video'ing.

And yet again, multi-benchmarking, as in, how you multi-task at home running a few things at the same time and not just one application on it's own with Explorer.exe closed () is supposed to be where the IMC and Phenom shines more than single application benchmarks (check PCMark Vantage scores). I have benchmarked this but not seen most others do it or mention it (run a few different tasks at the same time, i.e virus scan, game, video decode, compression, integer calculation and some encryption and test).

That bug right there sounds a lot like Errata 254, possibly 298, which may explain why MS applied a software TLB fix in Vista SP1. So if the voltage delta between memory and IMC also causes an issue, then perhaps 1.8v memory was the way to go?

BSoD does sound similar to the errata. It won't be though, since he only get's it on multi oc, which is typical instability. Let him switch to XP or Linux and replicate it.

1.8V RAM? I'm not sure. Achim is currently testing settings I want to run and he has the same RAM as mine. We'll see how long his lasts, if it makes it past 3 months without death/degradation, then it can't be a problem on all Phenoms. If not, then it sure is and we would need to advise that very clearly. ATM I am very reluctant to tell anyone to run NB below 1.2V if RAM is plus 2.1V. Maybe this is the exact reason CnQ did not change NB volts/speeds, even though that actually does affect power consumption/temps very much?

In other news though, I got my new Toxic Edition 3870 installed now, with 8.3 drivers. Not have any of the issues I had with the other card, even while having the CPU OCed.

Nice. So have you tried oc'ing it yet using CCC or ATi Tool?

Hmmm finding the actual values for current, could be difficult since it can't be measured properly without making the Multimeter a part of the circuit.

Yeah, that was the initial problem I had. A Clamp Ammeter is OK, but still not as accurate as we want. AMD uses the Extech 380801-NIST Power Analyzer (you must have seen it in video demo's too, they recommend that), and that's deemed accurate for measuring this and logging this as the system plugs directly into it and various loads can be configured. Costs roughly $700 though and I haven't found it around here. The current chosen at boot on a Phenom is written in an MSR value, so you can find it and decode it using AMD BKDG. It differs. Mine was found to be 76A stock.

But it does make sense that the actual current in the die usage could be 67amps. After all the hard input to the cpu out of the PSU is 12v, and draws about 6.7 amps off the 12v rails. Considering a CPU staging area is a massive parallel circuit, which if set up right divides voltage, and after the voltage is divided, you could put a current amplifier circuit off of each branch. When I was doing electronic classes in such in college it wasn't difficult to set up an amp circuit that could take 200mA in and output 2Amps, granted in those circumstances the voltage wasn't lowered so soon after the resistors and transistors would start to let the magic operating smoke out.

That's the problem. IMC is a big power/heat hog, watch Nehalem at 32nm and compare it to 32nm AMD. Then we'll be talking apple to apple comparison. HT/MEM is controlled all within the CPU of K10h. Intel has this in it's Northbridge, and usually a 35-50W NB that is. Add that in and you'll have the power consumption of both CPU's working. But you would not need to use a modified ammeter to measure such a load, as a CPU does not pull all it's load from the 4/8-pin +12V but also through the MB +12V rails. Many of the SMPS out there run +12V rail off the same rail as is used for the MB 24-pin connector, so you can't really track individual component power down. You would need to account for that. Best would be, if you could verify all power, real idle/load I.V for all the various components with and without the CPU. Then you'd have a much more accurate power measurement.

Than you have CPU Pd and Ps, static and dynamic load (differ).
Further you have multi-tasking loads (again, differ).

@KTE

thanks, if there is some "expermental" way to fix this crap :P, let me know, i'll spend what time i have to test stuff if you think of a way to attempt to "fix"

im any case it looks like i'm about to buy some ram, don't feel right running one stick with 1/2 heatspreader (and only one heatpipe)
soo, the 1.33 bios is good for 1066?

Try P0J BIOS please and see if those options are fixed. At first, only leave one RAM module in.

When you flash BETA BIOSes, ascertain a minimum damage and maximum probability approach, which is to keep minimum components attached, i.e. 1xHD, 1xRAM, etc After first boot, Load Optimized settings, reboot, in BIOS, you can then attach all components back and startup with the correct settings you want.

I will try to check your issue out dr_drache, but I need a Phenom to do it first.

thanks KTE,

I am using 2* 1GB Modules HyperX PC8500 from kingston. I installed the ram into the first 2 slots, orange and green nearest to the cpu :P ( knew the color coding was wrong from a review ). I am using the official 1.3 bios. under CPU-Z, my timings are:
Memory Type DDR2
Memory Size 2048 MBytes
Memory Frequency 533.4 MHz (3:8)
CAS# Latency (tCL) 5.0 clocks
RAS# to CAS# (tRCD) 5 clocks
RAS# Precharge (tRP) 5 clocks
Cycle Time (tRAS) 15 clocks
Bank Cycle Time (tRC) 14 clocks
Command Rate (CR) 2T

Your RAM is rated for

Clock Cycle Time (tCK) CL=5 2.5ns (min.) / 8ns (max.)
Row Cycle Time (tRC) 51.5ns (min.)
Refresh to Active/Refresh Command Time (tRFC) 105ns
Row Active Time (tRAS) 39ns (min.) / 70,000ns (max.)
Power 2.025 W (operating per module)
Operating Temperature 0o C to 55o C

Keep tRFC as it mentions for DIMM0 and DIMM1 and set the volts as 2.2V (users: check the specified power draw of each module for reference)

Load test it with Memtest (latest). Does it pass at least 4 hours?

C'mon KTE... 9850 BE's are only a couple weeks away.... Is she a go'er know what I mean, nudge, nudge, .... Uuaaa....

http://www.xtremesystems.org/forums/...d.php?t=180276

Just what I need, another Phenom, but I'll probably get one it the price is reasonable...

It would surpise you how little computing interest I really have if you met me or were around me for a day or two.

That isn't to be released till ~10th April for us and it's at 125W TDP B3 for 2.5GHz (customer deliveries). Only OEM and shipments to retailers are now in March.

9150e 1.8G 65W
9550 2.2G
9650 2.3G
9750 2.4G 95W and 125W
9850 2.5G 125W

Two higher bins follow, one is 9950 in Q3 and one more is Deneb 95W in mid-September.
Two bins of FX also follow, one TBA this 2nd half and launched with 9950.
B3 Barcelona's will be upto 75W for sub 2.3G
B3 Barcelona 2.5G will be 105W rated

I might be able to get the B3... I'll check by the uncles tomorrow. If I can, then I might as well test it and if not for free (meaning I have to buy it like a normal user), then I won't be.

Also maybe there is still some confusion with enthusiasts about the TLB bug.. I'll make it clear. The Barcelona B2 server parts were withdrawn because the TLB Errata gave many problems in the server world at the very high multi-core loads. It was affecting many initial release systems. But there were no problems in the desktop world so these were carried on.

KTE, AMD started to use a new nomenclature for dissipating power in phenom CPU's...its called ACP(Average CPU Power).Real TDP should be around 120W.

Yep, I'm aware. But that was mainly server world, i.e.



Using the above power meter I listed, AMD tested SPEC dynamic and static full loads, through the highest benchmarks not available to be replicated by consumers. The highest loads they observed at any 1 second period with the hottest bin in each SKU from the fabrication line is what they set the TDP as, as that is the maximum heat-power dissipated by the chip. Even those they made clear were not attainable under any maximum load outside of fab. That's what decided Phenom TDPs as we know.
The reason they did this was as the theoretical peak TDP was never observed in real-life usage, ever. The peak everyone was interested in is what could be observed with the maximum software load available for their home desktop/workstation/server. Mostly, server departments would moan about having many server racks free but not enough power available to add extra systems (thereotical data and calculations are for their assistance primarily). Their facilities manager would tell them, you cannot add a single Watt more. So they would moan and ask about how they could decrease this wattage and yet increase performance and fill the racks. That's why AMD then tested to give out the maximum power the server can pull out of these CPU's at full load, as that would then tell you better what cooling and power supply you would need and the bills per month for that entire rack.

AMD then publically stated to everyone (non-verbatim): "Buy a $30 domestic power meter and run your system with the intended software, whether Intel or AMD, at full load. That is the way to get your power, not by looking at TDP, which is worst case scenario intended for cooling solutions and thermal management only." (search John Fruehe comments)

As in home, P95 v25.6 didn't even get close to 93W for my Phenom. The way I'll measure it next is using the Gigabyte Odin PSU logger, it gives you the amps draw very close to fully accurate from each rail. Once I can tap off which rail is powering which connector, I can get the amps and measure the volts using a DMM and through the software to ascertain maximum stock idle and load power by the CPU and components.

Look to Achim too, he was after getting the same utility and PSU.

As for power meter results: I'l post some later on.

=========

A side note. Check this, as I stated, a paper hyped launch. My boss (UK) was after the SkullTrail for a long time... no availability here for at least 3 weeks yet (Intel says directly). Prices:
$4100 = $2050 x 2 for both QX9775
$800 for MB
$2000 for CAS5 ValueRAM 4x 4GB

And no stock on any of those yet (check Tekheads for instance). Absolute lunatic prices. I would rather by a 16-core Tigerton and annhilate that performance in half.

[Mathos]

@KTE I think you mentioned in another thread that MSI is working on a new bios to add NB multi and the settings for P-states back into a newer bios. Yesterday they finally replied to my online customer service request regarding the same thing. They gave me a username and PW but it keeps giving the alert 530 not logged in when I try and use it there to get to their private bios site. So soon as I get that working I'll let you know what they've got set up. Specifically asked for a cross between 1.1b3 and P0j

And cmon, stay with the dark side....you know you wanna do it ;P. Only thing I can recommend is to not use the stock cooler this time. Nirvana 120 premium, TRUE, or something of that caliber, should easily handle the heat.

On the other hand, the multiple cpu death thing that seems to revolve around the NB/IMC kinda gives hard evidence of a hotspot on the die causing probably the TLB and part of the reason the phenom were released with the IMC/L3 being clocked so low.

And as far as all that can be tweaked thats why I enjoy mine so much as well, granted *knock on wood* I haven't had the processor death problem yet. Granted I've also been able to OC at lower voltages than you have as far as I can tell though as well which probably helps. I kinda wish those ODIN Psu's didn't cost so much.

I think I'm actually about the same way when it comes to showing interest in computing in reality. Now IF I'm talking to another person that's interested, then yeah I'll talk about it. But same here it's by no means my life. Though I've actually recently started to play games on my computer again, which is something I hadn't done much of since I got the phenom build goin. I gotta laugh, I've been the same way about roleplaying gaming for a long time as well. I enjoy RPGing, but I don't talk about it outside of the game, or in the rare occasion it comes up in conversation, even though thats about a 50/50 whether I'll talk about it. Bad part is I've had friends on the other hand, that do nothing but talk about gaming 2/3 of the time, thats what kinda turned me off to being such a dork about it.

And no I haven't tried to OC the Toxic edition yet, but let me tell ya the VaporX cooler is freakin amazing. Card Idols at around 40-45c, at that temp the fan rarely runs if at all, full load in half life 2 episode 2 with high setting it stays below 60c, and even then the fan isn't going above 50&#37;, and thats without ATiTool, if I use ATiTOol it runs even cooler. Thats at factory default 800mhz core, should easily be able to go much faster, but I'm a noob when it comes to OCing vid cards so not sure how high I could get, should hit 825 easy, since that was what the atomic which is the same card was clocked at, only difference between the two is the name on the card, and package contents. You don't get the $30 etailer thing, or the metal briefcase, you get a 6ft hdmi cable instead of a 9ft, you get valve blackbox free (can't remember if atomic was orange box or blackbox.

[d412k5t412]

New version of AMD Overdrive!
2.0.17
http://www.techpowerup.com/downloads/1012/AMD_Overdrive_2.0.17.html

[Daveburt714]

New version of AMD Overdrive!
2.0.17
http://www.techpowerup.com/downloads/1012/AMD_Overdrive_2.0.17.html

I've tried this quite a few times and I keep getting the message that
"AMD OverDrive cannot detect AMD CPU on this computer"...

Am I the only one with this problem?

[aGeoM]

I've tried this quite a few times and I keep getting the message that
"AMD OverDrive cannot detect AMD CPU on this computer"...

Am I the only one with this problem?

No, same here. DFI 790FX board / Vista x64.

[d412k5t412]

Works for me, but I had a fresh install of XP too. Try that if you're that desperate? Couldn't auto-clock passed 2.4ghz