Hello all,
Last week I got a new Athlon X2 model processor,
which is for OEM only and have a TDP of JUST 22W!
I tested it a bit and below are some pics for your reference.
Printable View
Hello all,
Last week I got a new Athlon X2 model processor,
which is for OEM only and have a TDP of JUST 22W!
I tested it a bit and below are some pics for your reference.
Nice. Technically an 11Watts per core CPU. How is the overal performance? And how does it OC? lol. Any benchmarks? Is it passivly cooled?
22W is nice, mostly for HTPC builds...but 1500MHz? :/
I've been looking for this but apparently it was slated only to be sold to oem's not sold as oem.
it's meant to go head to head with atom, 1500 is all it needsQuote:
Originally Posted by Morais
yup, right now I am testing the CPU with the best ITX board - MINIX 780G SP-128MB.Quote:
Originally Posted by Morais
I am almost done and will upload more results soon, stay tuned! :cool:
revision G2 ?
i found most AMD G2 processor's can run 2,8 Ghz plus .
i think this processor can run 2 Ghz + .
you're absolutely correct, stay tuned!Quote:
Originally Posted by keyboard
default vcore 300 HTT
http://i82.photobucket.com/albums/j2...core300HTT.jpg
default vcore oc
http://i82.photobucket.com/albums/j2...ultvcoreoc.jpg
default vcore oc, more
http://i82.photobucket.com/albums/j2...2GHzorthos.jpg
default vcore oc, more and more
http://i82.photobucket.com/albums/j2...1GHzorthos.jpg
I still have more to come up, stay tuned ~
2.3GHz done, still - default vcore
http://i82.photobucket.com/albums/j2...c23ghzlive.jpg
If you love my DIY case, please let me know :D
http://i82.photobucket.com/albums/j2...n/DSCF0984.jpg
It should perform a lot better than ATOM with 1.5GHz. :DQuote:
Originally Posted by chew*
yah I figured it would now they just need to make it available to the public, I wanted to build one with a 780G and if my math is right the platform will draw 50w max........Amd needs to release this to the public, or not bother competing with atom, I will not buy a prebuilt rig with this cpu.....I want to set it up for my needs, not what someone thinks my needs are. You have a killawatt there? Would be nice to know if my math is correct for 780G.Quote:
Originally Posted by kenofstephen
I don't have the meter, but it should be <60W if I don't oc.Quote:
Originally Posted by chew*
I use this board:Quote:
Originally Posted by chew*
http://www.jwele.com/UserFiles/jwele.com/1217225097.jpg
I'm still using an Athlon XP Barton core laptop, so any chance of upgrading to an AFFORDABLE dual core K8 @1.5 GHz laptop with decent on board graphic is certainly appreciated by me.
that would be sweet if they sold a board like that
cool case fan, isn't it? ;)
http://i82.photobucket.com/albums/j2...n/DSCF0985.jpg
This CPU is fantastic!
I must try my X2 4450e at 1.075V/2.3GHz (default is 1.25V). Wonder if it will work ...
With only 22W TDP and possibility of passive cooling this is perfect for many niche applications like CarPC, ULPC, industrial PC and more.
Thanks for sharing kenofstephen!
You're welcome. :)Quote:
Originally Posted by Lightman
What psu did you have in mind while doing the math? The foxtronn psu they used here gives quite impressive results.Quote:
Originally Posted by chew*
For comparison an setup with an GBT 780g S2H board an 4850e, 2GB DDR2 800 mem at 1.8V and 160GB sata2 hd and an seasonic II 330W psu requires ~39W AC in idle and ~85W AC with worst case load (CoreDamage).
Should be round ~30W/77W AC with the fortron. Doing an rough estimation now and simply using (77W-30W)/2 as the delta results in an 30W/53,5W AC consumption.
Close to your estimantion and without taking lower mem and board requirements into account.
I use thos SilenX on my wc setup, really very silent, I like em too.Quote:
Originally Posted by kenofstephen
+1 for not using the word HTPC here. :up: :rofl:Quote:
Originally Posted by Lightman
oh ya, you need to try higher than that ;) :p:Quote:
Originally Posted by Lightman
this mini-ITX board is so good at HTT overclocking! wow!
http://i82.photobucket.com/albums/j2...oreoc24GHz.jpg
I think G2 rev will OC up to 3GHz+
Can some test with another motherboard with higher HTT? For max clock this CPU :)
I don't prefer adding vcore as the major selling point of this CPU is "energy-saving", "low-power".Quote:
Originally Posted by bondhahnmrt85
@ default vcore oc to 2.4GHz one of the cores become unstable.
I know adding a little bit vcore can fix it but I don't prefer doing so. :)
not really, this ITX board can do over 400 HTT~...Quote:
Originally Posted by bondhahnmrt85
here you go~ 440 x 5~
http://i82.photobucket.com/albums/j2...tephen/440.png
nice:clap:
Wasn't really bringing a psu in to the math but it would have to have a high efficiency coupled with low watts. I had found a link to power consumption of boards hold on its in the original thread about this cpu, I will link it.Quote:
Originally Posted by justapost
The reccomended chipset for under 45w consumption with this chip is 740 advertised here http://www.xtremesystems.org/forums/...ighlight=3250e
If we look at this chart we can see that a 780G platform consumes only a few more watts than 740G http://www.silentpcreview.com/article859-page5.html
I rounded it up to 50 watts. The small cost of watts is worth it for the better performing IGP imo.
http://www.newegg.com/Product/Produc...ame=AMD%20740G
average cost of 740G $56.00 with hd2100 graphics.
http://www.newegg.com/Product/Produc...ame=AMD%20780G
average cost of 780G 79.85 with hd3200 graphics + better southbridge + wider variety of board options.
Worth the $25 difference IMO and what amd should be pushing for with this UVC line of cpu's. This Psu would fit the bill. http://www.newegg.com/Product/Produc...82E16817151059
GBT will ship new mATX boards with 780/SB750 also newer 740g boards, may be worth waiting.Quote:
Originally Posted by chew*
Thanks for the silentpcreview link, looks interesting but need to sleep now. Only had 1hr yesterday. Only saw they used an laptop hd.
Wonder if its possible to tie with a atom in power usage now and get much higher performance at the same time.
If I get this and a HD4550, is there then some fancy idle features so that gfx dont burn a lot of watts while not doing anything? Because then I should be able to get a very cheap running HTPC
It does not meet amd's specs for a smooth HTPC experience. This is geared more towards low power consumption/longer battery life netbooks. IIRC correctly the 740g setup with the 3250e is only 5 watts over an atom platform, so 780g would be 10 watts at most.
Well shouldnt the hd4550 be capeable of doing all the decoding? Aka meaning the cpu is kinda meaningless?
I would asume AMD has there reasons, it could be they overspec it so that no users are complaining about non fluid dvd/blu ray playback.Quote:
Originally Posted by Mech0z
decrypting DRM :banana::banana::banana::banana::banana:?Quote:
Originally Posted by Mech0z
Xbitlabs did a review on the HD4550 that could give the answers you are looking for.Quote:
Originally Posted by Mech0z
http://www.xbitlabs.com/articles/vid...on-hd4550.html
Really cool. I wonder how much power usage drops when its under volted from stock. ;)
WOW~ this CPU and 780G ITX board are excellent choice for my new HD ITX project~ Great~
Thanks for sharing your experiences Kenofstephen. I'm due to upgrade my PC sometime in '09 and I've been thinking of going small and air cooled this time. Your OC results with that J&W board are truly impressive. Now if only it had a 8x or full 16 lanes of PCIE... That'd make my decision AMD fo sho.
agreed , since both of chip ( 3250e & 780g ) used less power , it would be true if it only have 10w ~ 15w power consumption .:up:Quote:
Originally Posted by chew*
Do you have Sempron x2 2100+ too ?Quote:
Originally Posted by kenofstephen
It would be interesting to compare Sempron x2 with Athlon x2 22W :welcome::welcome:
sseriously guys, you don't really need this exact CPU if it's hard to get hold of, the real-world results of making your own out of any G2 will be similar :)
I've had G2's 2 1.5ghz mark and 0.8v running fairly stable without a heatsink at all. just some airflow over the ihs
Ran a dual-core cinebench run.. peaked at 100c no crash :lol2:
are you sure ?Quote:
Originally Posted by mAJORD
that's cool :cool:
the processor not broken ? :eek:
no it's fine..
I was going to make a vid of it as a demo of how power efficent you can make them.
unfortunately i do not have a watt meter
but i am able to run my phenom 9550 @ 1.5ghz @ .8v full prime temps top at about 31C (xigmatek 92mm on low fan)
Quote:
Originally Posted by Mech0z
Look at it like this: Is it possible that AMD K8 architecture @ 65nm has better performance/watt than Intel architecture which is designed for low TDP applications and is manufactured at 45nm?
Simple answer: No.
Simple answer: yes, it's POSSIBLE.Quote:
Originally Posted by Calmatory
I think it's possible because Atom uses old, power hungry 945G chipset comparing to other chipsets.Quote:
Originally Posted by Calmatory
Do you understand that this AMD CPU is 1.5 GHz and has TDP of 22 W, while Intel Atom has TDP of less than 4 W? Dual Core Atoms will have TDP of 8 W.Quote:
Originally Posted by knopflerbruce
You do the math. :rolleyes:
Want to count the chipset in? 6W for the 945GSE chipset. :) So, 10 Watts for the CPU AND a chipset. Oh, that is less than 50 % of what AMD CPU takes alone. :up:
Still possible? :ROTF:
your math is off and it's already been tested.........atom platform does like 5 watts less........than the single core amd variant on a 740g.......ATOM also has a weakness, I forget what it was but the best way to decribe it is a one way street. In the watt department its a win, performance its a fail. I'll take the higher performance at the cost of 5-10 watts.Quote:
Originally Posted by Calmatory
TDP for AMD is HEAT out putt at max load.Quote:
Originally Posted by Calmatory
TDP of 140 9950 consumes 100 watts max load.
How is my math off?Quote:
Originally Posted by chew*
But wasn't the comparison about perf/watt, and not highest performance? :rolleyes: If you want more performance, take a Wolfdale. ;)
AMD CPU has two cores though, which is a clear win, but once the dual core Atoms hit the street, AMD will lose that advantage aswell. Even single core 1.5 GHz AMD can't compete against dual core Atom power consumption wise unless AMD moves to 45nm.
Atoms can idle @ < 1 W and they take less than 4 W under load. I doubt this AMD CPU can idle at less than 4 W.
TDP stands for Thermal Development Power, and is mainly aimed at cooler makers whom job is it to make sure that the cooler is powerful enough to dissipate the TDP amount of heat in such way that the CPU does not overheat.Quote:
Originally Posted by demonkevy666
100 watts based on what? I am just writing an benchmark so could you define a "max load" for me and tell me how to get CPU to that state? That could make me rich!
unfortunately for atom it cant sit on your desk by itself so you can surf the internet, instead it sits in a motherboard that it comes with........that does consume wattage........can't compare the cpu/cpu when they can't operate on they're own....in fact they only sell it in a motherboard so that now becomes part of the equasion. Totally sidestepped the one way traffic fatal flaw of ATOM btw, thought i would point that out ;)Quote:
Originally Posted by Calmatory
Same goes for amd, they package it with a motherboard, or oem's do and do not sell it by itself so when they are available thats how they will have to be compared.
Quote:
Originally Posted by knopflerbruce
agreed :yepp:Quote:
Originally Posted by BlueSkyNIS
read this :Quote:
Originally Posted by Calmatory
http://www.guru3d.com/article/ecs-at...p945gc-review/Quote:
The platform completely maxed out merely utilizes 55 Watts
:shakes:
...which includes motherboard components, RAM, Chipset, peripherals, hard drive/SSD/optical drive, PSU inefficiency and GPU. :clap: In other words: The whole system power consumption, not just the CPU+chipset.Quote:
Originally Posted by keyboard
The fact is that 945G SE Chipset + Dual core Atom have summed TDP less than 14W, no matter how you try to turn, twist or flip it. The AMD CPU? 22 W + Chipset.
IF This 22 W AMD CPU was SO great, why it isn't used for netbooks? Or, if AMD's K8 architecture scales so well @ 65nm, how does Intel's C2D Wolfdales do @ 45nm? :p:
There is a reason why Intel made Atom instead of ULV Wolfdales. ;)
i just got a single core atom @ fry's with a 945 mobo for $60 when do these go on sale cause i might just return this and get the AMD version performance wise watt for watt it will smoke the intel setup not to mention this WILL decode HD video 1080P no issues at all while the intel will cry w/o a VGA card which kills the whole low watt idea.
Oh, it is about HTPC now.... :rolleyes:Quote:
Originally Posted by cdawall
It uses onboard video, and the rest are requirements to run a system......it's not like they are comparing the amd cpu without all that :banana::banana::banana::banana: to a ATOM with all of it............bottom line the amd platform draws just 5 more watts and performs better.Quote:
Originally Posted by Calmatory
Once you put it in a platform its wattage consumption mounts to a hill of beans as well.........Intel took the low power cpu approach totally dismissing the platform as a whole, Amd took the low power as an entire platform approach which is smarter because you can't surf the inet via a chip sitting on your desk.........
You don't see them in netbooks yet because they just became available, we shall see in the next 6 months.
This discussion is over.......
bah just looked into the jetway board $150 is a bit much considering how cheap the intel stuff is....Quote:
Originally Posted by Calmatory
You don't see them in netbooks because 22W is far too much power to cool within a netbook. Current single-core Atoms netbooks use around 12W in total at full CPU load.Quote:
Originally Posted by chew*
So 330 dual core atom consumes 14-15W?
http://www.abload.de/img/lpha6l.jpg
Complete system with 4GB ram an HD2400pro, Hitachi 160GB 7200rpm SataII HD and totaly oversized watercooling consumes:
Idle: 67-68W AC
Load: 80W AC
So 12W difference with small fft's.
Abit unfair because I use the lowest possible voltage.
If I add 0.01V using 0.95V now
http://www.abload.de/img/lp2mlle.jpg
Idle: 68W AC
Load 82,5W AC
Difference 14,5W AC.
CoreDamage pushes IES to 21W but total system consumes 81.1-81.6W. Substract 30W from system consumption, because lowest I could reach with an 4850e was round 38W AC.
Few benchmarks for comparison with guru3d's 330 review
http://www.abload.de/img/guru3dcomp616i.jpg
Drystone:
atom 330: 7804
4850e(1.5GHz): 8456
Mandel:
atom 330: 500
4850e(1.5GHz): 1401
At least for me, this is a closed case.
A complete netbook using a dual core Atom would use around 14-15W under full CPU load.Quote:
Originally Posted by justapost
The point of Atom is a dirt-cheap CPU that has low enough of energy use to be put into a tiny and inexpensive package without worrying too much about cooling and whose performance is adequate for typical usage. Intel already has the unmatched C2D for any other mobile application where performance is of any importance whatsoever.Quote:
At least for me, this is a closed case.
Ah ok, guru3d showed a total difference of 8W. Looking at mandel results it's still a closed case. With a 220W fortronn psu the total consumption of an X23250e system will be sub 50W.Quote:
Originally Posted by accord99
Netbooks? OT. :)
Hmm miniITX board + 330 is in 70€-90€ price range here, do you really think 3250e+board will be more expensive?Quote:
Originally Posted by accord99
I thinks it's similar to i7 vs. PI/C2/(PII ?). The cpu consumes more power under load but does the job quicker and ending up more energy effective overall.
Uh... because you don't need anything more powerfull than an atom to run a browser or play solitaire?Quote:
Originally Posted by Calmatory
and that reason is because it's more cost effective.Quote:
Originally Posted by Calmatory
Current AMD mini-itx motherboards aren't that cheap and they're already too large for netbooks. Besides, I don't think AMD really wants to get into this low-margin market.Quote:
Originally Posted by justapost
Perhaps, but the mobiles, and netbooks in particular, constrain how much cooling you can put in.Quote:
I thinks it's similar to i7 vs. PI/C2/(PII ?). The cpu consumes more power under load but does the job quicker and ending up more energy effective overall.
Maybe for the netbook market but for the desktop market for average joe looking for sale items on best buy, sending an email, sharing pics with family over the net.........It's a valid market.....Its the only comparable desktop that completely sold out at best buys within 300 miles of me over the holiday season.......1650le's, 1 gig ram,nvidia chipset/onboard graphics with 160gb drives...........oddly enough I could still buy an atom netbook right up untill 6pm when everything closed at a bestbuy 5 miles from my house.
never mind....
and...how much does this cpu cost?
How much will it cost us or the OEM's. It's only being sold to oems, not to public......so it appears you will have to buy a barebones at the very least.Quote:
Originally Posted by SupeRClockeR
Justapost: Measuring wattage with software? .... :(
Original question was: "Is this better than Atom?". Is it? (Note: The thread is about CPU, the question was about the CPU, however someone wanted to take motherboards and platform in aswell.)
So, if one asks whether or not this is better than Atom, what would people answer? I remind, AMD ULV Desktop CPU with (Assuming it is a ULV Brisbane) 154M transistors @ 65nm, Intel ULV UMPC CPU with (around, 47M per core)95M transistors @ 45nm. The comparison is twofold: Desktop CPU vs UMPC CPU, other one shines in performance and other one shines in low power consumption. What matters the most? For desktops(in this case HTPC's aswell) the performance, and for UMPC(netbooks for example) applications the low power consumption. :shrug:
The absolute facts: 3250e is faster. Atom is more energy efficient. There is no absolute answer/truth.
What really interesting is that it fits in the standard AM2 940 pin socket. The Atom requires you to use their NB/SB arrangement.Quote:
Originally Posted by chew*
Interestingly, Intel is currently trying to prevent Nvidia from using Atom with a downscaled chipset they did called ION.
This could get interesting
These "TDP" numbers and arguments are all pretty bogus, totally splitting hairs. Anything below 25-30 watts is pretty good in general for these types of devices. Atoms also have SEVERE limitations which have caused their adoption to be stymied to some extent.Quote:
Originally Posted by Calmatory
TDP numbers are the newest thing for "expert" review sites to talk about, but at those levels, it doesn't matter that much, unless you're doing a super low power battery based state device, which Atom is not really being used for (this is the ARM market)
The 10 watt difference between the 3250e and the Atom 330 is well worth the difference for those building a standard device and want to choose which components to use. Atom based stuff is NOT the standard PC market....
You said it. ARM. Why wouldn't Intel want to get into ARM market with x86 CPU? That would allow people to run Windows and other x86 software on embedded devices, which currently run quite exclusively on ARM based CPU's.Quote:
Originally Posted by nanohead
I doubt this 3250e is trying to fight Atom really. Atom is also fighting VIA Nano, which is doing quite well. AMD also has to fight them.
As for TDP.. anything below 25-30W is good!? Have you ever used a device with a battery? :ROTF: What if the device is small, pocket sized?
No matter what the TDP is, Atom is way more energy efficient than what 3250e will ever be. <8 W vs.22 W.
The Atom as a CPU is indeed very power efficient, but take into account how power hungry the chipset is and how much superior the AMD offering is in both performance and power consumptions it's an easy decision for the desktop or embedded products as a platform. I'm not sure about you, but when I build a system, I select the components and take into consideration the platform as a whole, and not just the CPU alone. When it comes to power efficiency, and such low numbers we're discussing here, you need to consider the platform as a whole, not just the processor that goes in it.Quote:
Originally Posted by Calmatory
The only big downside to the AMD platform that I see is the cost compared to the Intel offering.
I just won a 2650e on ebay, cost me $50 (!):) My mom needs a pc for browsing (i hate it when she mess around with mine lol), this one should be quiet too. But $50..... OMG, I was expecting like half that price:p:
that's cheap enough .Quote:
Originally Posted by knopflerbruce
it's dual core isn't it ?
i must pay 43 $ for 3500+ single core down here .
:rolleyes:
2650e is single core, 1.6GHz and 512KB L2
3250e is a dual core, 1.5GHz and 512KB L2 per core.
I mesure with an power meter and use IES in addition. Tool is uesd to controll phases so it reads from a sensor at the voltage regulator. Previous measurements have shown it's quite accurate. For phenoms it must be taken into consideration that this tool does not measure the separate nb phase.Quote:
Originally Posted by Calmatory
Guy's on HT4U measured NB consumption on that phase with an modified board and an clamp meter. Was round 12-15W idle and 15W under load.
They had also measured the other four phases. I compared my IES results with them and they where quite accurate.
So I assume IES reading is good. I do not trust software readings in general. ;)
some more underclocking action to stir with :)
http://img518.imageshack.us/img518/6...osnofanjf3.jpg
http://img518.imageshack.us/img518/7...150008vvt6.jpg
And a Video of heatsink-less cinebench run as mentioned earlier
http://au.youtube.com/watch?v=qnZz28L_R48
Enjoy
WOW! :rolleyes:
This is great to watch! :clap:
BTW soon we will get platforms from AMD capable of running within 25W envelope! These will be more of a competitor to Intel Atom and ULV Penryns.
Want to know more??
Off topic Majord but how is the jigabyte board treating you?
Which mobo allows such low voltages? :eek:Quote:
Originally Posted by mAJORD
Since when have thermal sensors in Brisbanes worked? I.e. CoreTemp readings are wrong.
You groundbreaker, mine freezes at 0.8V :p:Quote:
Originally Posted by mAJORD
Also had an CubeCase running whit an 4850e and an HSF without fan. CPU temp (not core temp) maxed out at 50-55°C. CPU temps are normaly above the real temps. The sensors is onboard somewherennear or beyond the socket so he is also affected if the pwm area runs hot. Tested it like that: Placed a fan ower the pwm area and cpu+pwm sensor readings dropped. CPU itself was on water cooling.
0.6V is lowest possible on asrock. :)Quote:
Originally Posted by Calmatory
Recently I got information that *edit: 65nm* AM2 K8s DTS under reports by 21C. This comes directly from AMD as a patch for lmsensors in Linux.Quote:
Originally Posted by Calmatory
Jigga whats its name Jigga my Gigga dun dun dun dun........ sorry everytime i think of gigabyte i think of Jay zQuote:
Originally Posted by Calmatory
790gx gigabyte board
Those guys at Guru3D uses a 1000 W PSU, brilliant. :rofl: Imagine the efficency at 3 % load.
Here's another 330 system review, this one have a PicoPSU.
So instead of 50/52 C, 71/73 C are the accurate temps? :)Quote:
Originally Posted by The Coolest
I kinda find that hard to believe, if 21c is true for ALL K8 CPUs I've had temps at 100c during alot of air sessions... and when i kill chips it's usually not when they're in the socket:p:Quote:
Originally Posted by Calmatory
I'd like to have some proof... To my eyes these sensors were never that much off. My 2 856's had idletemps at 22c and 25c at stock in my server today, ambient was 11-12c (didnt close the window before I left for my x-mas holidays). Idling at ~35c above ambient at stock voltage and speed with GOOD coolers (blue orb II and the stock opteron cooler with LOTS of copper) seems like x-files to me.
I believe it applies to Brisbanes.Quote:
Originally Posted by knopflerbruce
I find it hard to believe aswell. Brisbane idles at 14-19C, while HWMonitor shows 34C. When HWMonitor shows around 55-60, CT shows the same. When HWM shows 80C, CT shows around 100C or so. :shrug:
However, if it does not apply for brisbanes.. I ran Windsor @ 110C both cores, and thats what CT showed. Add 21C and you got 131C. :cool:
Note: CPU's don't burn so easily really, they crash first. ;)
It's OK, i've ahd no failures, or cold boot issues, but it doesn't overclock my Phenom 1 very well. Overpriced IMO. for a budget board i'm more impressed with its baby 780G brother, MA78-DS3HP, much cheaper but OC's pretty well.. all it needs is better pwm cooling. Also lacks sb750 in exh for sb700. But no ones that fussed about the 750 anymore :pQuote:
Originally Posted by chew*
Quote:
Originally Posted by Calmatory
I think you'll find this is fixed in G2 Stepping, OR coretemp is taking it into consideration in this version (0.99)Quote:
Originally Posted by The Coolest
I am confident of this now as when it reports say 40c at idle, the actual heat spreader is only luke warm to touch.. At idle power at such a low voltage the thermal resistance between the ish top and the chip itself would be totally negligable, so in this case IHS temp = core temp minus say 1 or 2c at most.
You're right, I forgot to mention it was for 65nm AM2 K8 only, I'll try to see if it's G1 only, but from my understanding the offset applies to all rev G chips.Quote:
Originally Posted by Calmatory
Quote:
Originally Posted by The Coolest
I think this is the patch you mentioned.Code:Current Temperature for K8 RevG desktop CPUs is a "normalized value"
which can be below ambient temperature.
As a consequence lots of RevG systems report temperatures like:
sensors
k8temp-pci-00c3
Adapter: PCI adapter
Core0 Temp:
+17C
Core0 Temp:
+3C
Core1 Temp:
+21C
Core1 Temp:
+5C
being quite below ambient temperature.
There are even reports of negative temperature values.
This patch corrects the temperature reporting of k8temp for
RevG desktop CPUs.
Cc: Rudolf Marek <r.marek@assembler.cz>
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
drivers/hwmon/k8temp.c | 17 +++++++++++++++--
1 files changed, 15 insertions(+), 2 deletions(-)
diff --git a/drivers/hwmon/k8temp.c b/drivers/hwmon/k8temp.c
index ff12281..b138f79 100644
--- a/drivers/hwmon/k8temp.c
+++ b/drivers/hwmon/k8temp.c
@@ -49,6 +49,7 @@ struct k8temp_data {
u8 sensorsp; /* sensor presence bits - SEL_CORE & SEL_PLACE */
u32 temp[2][2]; /* core, place */
u8 swap_core_select; /* meaning of SEL_CORE is inverted */
+ u32 temp_offset;
};
static struct k8temp_data *k8temp_update_device(struct device *dev)
@@ -116,13 +117,15 @@ static ssize_t show_temp(struct device *dev,
to_sensor_dev_attr_2(devattr);
int core = attr->nr;
int place = attr->index;
+ int temp;
struct k8temp_data *data = k8temp_update_device(dev);
if (data->swap_core_select)
core = core ? 0 : 1;
- return sprintf(buf, "%d\n",
- TEMP_FROM_REG(data->temp[core][place]));
+ temp = TEMP_FROM_REG(data->temp[core][place]) + data->temp_offset;
+
+ return sprintf(buf, "%d\n", temp);
}
/* core, place */
@@ -176,6 +179,16 @@ static int __devinit k8temp_probe(struct pci_dev *pdev,
"wrong - check erratum #141\n");
}
+ if (((model >= 0x68) && (model != 0xc1)) &&
+ !(model == 0x68) && !(model == 0x6c) &&
+ !(model == 0x7c))
+ /*
+ * RevG desktop CPUs (i.e. no socket S1G1 parts)
+ * need additional offset, otherwise reported
+ * temperature is below ambient temperature
+ */
+ data->temp_offset = 21000;
+
break;
}
These are the constrains for the +21°C offset. Comments mention revG but there are a few excludes in the if clause.Code:if (((model >= 0x68) && (model != 0xc1)) &&
+ !(model == 0x68) && !(model == 0x6c) &&
+ !(model == 0x7c))
Basically everything that is NOT AM2 RevG chip is excluded.
Thanks, already had found that model value represents the model part of the cpuid. But could not figure out the relation between cpu and model number.Quote:
Originally Posted by The Coolest
Sidenote:
Rudolf Marek had posted an modified k8temp with k10 support few month back. Tcontrol_max is 70°C but phenoms have this offset value which must be manualy tweaked for calibration purpose. I think this makes proper reading difficult on k10's.
http://www.abload.de/img/tctle2sb.jpg
The offset can be in the range +10°C - -56°C, so in the worst case the core temp shown is 10°C below the real temperature.
Good news!
These processors are coming to RETAIL!! :up:
Source: Techpowerup.com
Guess I will throw the old X2 3600+ Windsor to the trash bin and get X2 3250e instead(for low power server), just mere 50-60 euros. :yepp:
That chip is also suitable for an low power/budget virtual server. In opposite to intel all x2's have hardware virtualisation support. :rolleyes:
OMG, I just paid $50 for one... and they'll cost less than that brand new:ROTF: Oh noes:DQuote:
Originally Posted by Lightman
Thats good news, I've been chomping at the bit for one of these, It seems amd has been reading the forums and is listening to can the original plan of OEM only with these ;)Quote:
Originally Posted by Lightman
Another naughty boy
the 15W 2650e,
it is not as new as the 3250e (just released),
but it is good for fanless setup.
http://i82.photobucket.com/albums/j2.../DSCF1404r.jpg
http://i82.photobucket.com/albums/j2...phen/2650e.jpg
http://i82.photobucket.com/albums/j2...en/2650e-1.png
Sweet. I'll post my OC's here with that one once my chip arrives (I guess I have to wait 2-3 weeks, depending on how fast I fill the flat rate envelope my mate prefers when he ships my stuff to me - 15 chips in each package:p:).
I'm glad the G2's seem to take quite high HTT, I have a feeling 3.6-3.8 is the max limit of these chips on air (suicide), and then I need like 450-475mhz HTT;)
What case is that, kenofstephen? Is it a mini-ITX specific case?
I have a 5050e and I love it!
Please , OC it !Quote:
Originally Posted by kenofstephen
:yepp:
http://www.abload.de/img/lowy6wr.jpg
Lowest possible suicide here, VID is 0.765V. 0.75 does not get into windows.
http://valid.canardpc.com/show_oc.php?id=479765