Barcelona Opteron 2350(B1) arrived

Finally I've received s7e9h3n's Barcelona - Opteron 2350(B1) this morning:)

The images is partially masked as these are not mine, just borrowed.
http://222.151.153.254/c-board/file/..._x2_masked.jpg

This is my system, ASUS KFSN4-DRE.
I've just tried with 1 CPU
...this CPU cooler is too wider a bit, two of same coolers can't fit this board at the same time...
http://222.151.153.254/c-board/file/...RE_CoolJag.jpg

This is BIOS screen shot.
Split Power Plane(now it's called DDPM: Dual Dynamic Power Management) is supported
on this board, KFSN4-DRE...it looks working fine with north-bridge clock at 1.8GHz.
http://222.151.153.254/c-board/file/...50_BIOS_NB.jpg

I tried SuperPI 1M with 1 and 2 memory modules.
39.750s in 1 module operation,
38.968s in 2 modules operation,
so it looks working dual-channel mode in 2 modules operation.
CPU-Z shows Single for both case...Fracnk can improve it soon, I believe:yepp:
A small window located left side in these screen-shot is FELIX's test program.
Also FELIX is working for proper detection of memory access mode in his MemSet:)
http://222.151.153.254/c-board/file/...Single-chn.png
http://222.151.153.254/c-board/file/...Single-chn.png

Now, OC time...very mild though;)
I could reach 2.4G with default Vcore and normal air-cooling(no assist with air-conditioner).

SuperPI 1M
http://222.151.153.254/c-board/file/...t2350_2.4G.png

CineBench10
http://222.151.153.254/c-board/file/...t2350_2.4G.png

...stay tuned:)

Nice one mate....now we want 2.8ghz-3.5ghz please :up:

At last some oc'ing of these beauties. Keep it coming. Me too want 2.8 - 3Ghz!!

is the DDPM really the split power planes, or is it just for setting different voltages to each socket? and can you post a screenshot of the bios page with that information?

also, why does cpu-z show the HT link as 200 and 240? is that a cpu-z error, or worse?

thank you sir

It is 7 second faster in super pi with just 400mhz up:eek:
Awsome :D

yeah, the performance increases are way above normal scaling for 400mhz, for both spi and cinebench. the enigma of k10 continues...

kyosen.. give em ln2.. see if there any coldbug or not..

32sec for 2.4Ghz AMD and very slow ram is awesome! Also this is server stuff, desktop stuff would be faster.
IIRC 3Ghz K8 with same ram could do 32sec spi1m.

Quote:

---

Originally Posted by tictac

kyosen.. give em ln2.. see if there any coldbug or not..

---

I think it only needs a single stage to find out that..... :D

That is some strange scaling going on there:confused:
Looking forward to more tests!

Thanks Kyosen for bringing first serious Barcelona overclock!

can you run latency tests on stock and OC for comparision? THX!

PS. Can someone write here resoults from CinemaR10 picture coz it's not loading on my PDA...

Quote:

---

Originally Posted by Frodin

That is some strange scaling going on there:confused:
Looking forward to more tests!

---

The scaling is down to the cache/IMC that changes speed with memory timing and/or CPU speed. Well I think that's what I read somewhere.

So yes, the scaling of applications that use cache will show strange results.

Quote:

---

Originally Posted by hipro5

I think it only needs a single stage to find out that..... :D

---

X1900XTX and X1800XTX were coldbuggers and i remember people fearing the HD2900 bugging like hell, now people are complaining that they can't cool them :D

I hope the same will happen for K10 :D

Nice run kyosen, looking for more results.

superpi times remember me to A64 Venice @ 2.8Ghz or somewhere near it :)
hope it can scale much higher

thx for sharing :up:

Not bad...not bad at all.... 400MHz without touching the vCore sounds good, and the scaling is looking good, also...

Maybe now we can see the "good" side of the K10...:up:

Keep us updated...

Quote:

---

Originally Posted by wedfighter

superpi times remember me to A64 Venice @ 2.8Ghz or somewhere near it :)
hope it can scale much higher

thx for sharing :up:

---

2.6ghz actually could do the trick with BH-5 at high speeds..

Quote:

---

Originally Posted by BeardyMan

X1900XTX and X1800XTX were coldbuggers and i remember people fearing the HD2900 bugging like hell, now people are complaining that they can't cool them :D

I hope the same will happen for K10 :D

---

The 2900XT was designed by ATI and NOT AMD.... :p:

Quote:

---

Originally Posted by kyosen

...stay tuned:)

---

Good show so far Kyosen. I'll stay tuned. Keep tuning that rig. :D

Interesting results. I think we will be quite surprised when we can bench >3 Ghz Phenoms + oced memory.

And they said that K10 wouldn't overclock. Nice job, great scores. Waiting to see more great stuff. Thanks. :up:

At last someone is overclocking a K10, wipeeeeeee

that took toooo long coming now hopefully we will be able to get our fix, heeheheee

Looking forward to the results and the mysterious K10 scaling ;)

Pics removed coz its not a fair comparison.

Sorry!

K404

you seem to miss something ;)

K10 barca has L2 512Kb cache , L3 does nothing fro superpi.

Therefor the result is good at 2.4ghz :)

Just a quick look, you're 100 mhz faster and with much better memory timmings. Try running it at 2.4G at the same timmings as Kyosen, then post your score.

strap a 8800 gtx in there and post some scores :D


edit: 3dmark that is

Thank you so much Kyosen.:D
Seem i have to buy asus, it oc higher and seem run faster than tyan.

actually 32s is pretty good...i remember i scored around 35 with an 512k X2@2.4 with DDR2-800 and that each 200mhz increase would decrease the time by about 2s...so a 400mhz increase for k10 giving 7s difference is incredible.

L3 does nothing? damn. ok. Ill take those pics out, I was thinking about it anyway and 7 seconds for 400MHz gives more than 1 second for the clock difference, so theres 1-2 seconds (maybe more) gain in Barca clock for clock

I know this is probably not a valid comparison, but here's what I get on a Opty 185 with the best match I can do with Kyosen's timmigs. What I get at 2.4G and what it took to "match" his 32s.

with al due respect but your mem timings are screwing teh bench :D

http://www.johanlemaire.be/OC/A64%20...i%2032.890.JPG
http://img486.imageshack.us/my.php?image=superpi7mm.jpg

K10 and K8, it's 400mhz difference in superpi , but then again it might scale even better...

btw if

2.0-->2.4ghz = -7secs 32.XX secs
2.4-->2.8ghz = -7secs 25.XX secs
2.8-->3.0ghz = -3.5secs 21.5secs

That means it's roughly 8secs faster then K8 L2 512kb at 3.0ghz LOLOL

Kyosen,great results man :up: .And s7e9h3n is to be thanked for sending those chips :D.

Can you run Cinebench 10 in 64 bit OS?Also ,what are the temperatures of the K10s while OCed to 2.4Ghz,did you notice any jumps in load temps?

Good luck with them

Quote:

---

Originally Posted by informal

And Steven is to be thanked for sending those chips :D.

---

I didn't send anything :D

j/k lol

Quote:

---

Originally Posted by BeardyMan

with al due respect but your mem timings are screwing teh bench :D

http://www.johanlemaire.be/OC/A64%20...i%2032.890.JPG
http://img486.imageshack.us/my.php?image=superpi7mm.jpg

K10 and K8, it's 400mhz difference in superpi , but then again it might scale even better...

btw if

2.0-->2.4ghz = -7secs 32.XX secs
2.4-->2.8ghz = -7secs 25.XX secs
2.8-->3.0ghz = -3.5secs 21.5secs

That means it's roughly 8secs faster then K8 L2 512kb at 3.0ghz LOLOL

---

Yes, I know, I did that intentionally. Thats was the idea.

Quote:

---

Originally Posted by PhilDoc

Yes, I know, I did that intentionally. Thats was the idea.

---

yup my bad, misread that part :D

Thanks for benching.

Can you run 3dMark 01 to 06 at 2.0Ghz and at 2.4Ghz for us?

Quote:

---

Originally Posted by BeardyMan

with al due respect but your mem timings are screwing teh bench :D

http://www.johanlemaire.be/OC/A64%20...i%2032.890.JPG
http://img486.imageshack.us/my.php?image=superpi7mm.jpg

K10 and K8, it's 400mhz difference in superpi , but then again it might scale even better...

btw if

2.0-->2.4ghz = -7secs 32.XX secs
2.4-->2.8ghz = -7secs 25.XX secs
2.8-->3.0ghz = -3.5secs 21.5secs

That means it's roughly 8secs faster then K8 L2 512kb at 3.0ghz LOLOL

---

that's some very optimistic (wrong ;) ) math here.
I'll translate it in %:
+ 20% freq. = 18% lower time
+ 16,7% freq = 21,8% lower time
+7% freq = 14% lower time

edit: with wrong math I mean the scaling. at your scaling an K10 at 4,2GHz will have a 1m time of 0s

looks nice,dont forget this is only B1 stepping :D

Quote:

---

Originally Posted by BeardyMan

2.6ghz actually could do the trick with BH-5 at high speeds..

---

actually 2.7Ghz on Sempron Manila could do the trick :)

http://img3.freeimagehosting.net/uploads/2c80be5e51.jpg

surely K10 is slight faster than K8 on Superpi, hope it can score way higher on another benchmark

i really missed the AMD time ...... like what Venice and SanDiego did :up:

Quote:

---

Originally Posted by wedfighter

actually 2.7Ghz on Sempron Manila could do the trick :)

http://img3.freeimagehosting.net/uploads/2c80be5e51.jpg

surely K10 is slight faster than K8 on Superpi, hope it can score way higher on another benchmark

i really missed the AMD time ...... like what Venice and SanDiego did :up:

---

wut...you have your Kingston DDR2-533 running at 1080? how the hell did u do it?

Quote:

---

Originally Posted by wedfighter

surely K10 is slight faster than K8 on Superpi, hope it can score way higher on another benchmark

i really missed the AMD time ...... like what Venice and SanDiego did :up:

---

It seem to be. With the same L2 cache it took you 300 more mhz and 100 more fsb to match its score. Not to mention a little better memory timmings.

If the Phenom oc's as well as this Opty, then we should see people getting 3G on air with the first models. By the time the 3G Phenom comes out the AMD spi record will have fallen.

I'm glad Kyosen-san FINALLY received these cpu's. They were held in Japanese customs for 3 or so days. I've given him the go-ahead to run LN2 if he likes :D .

One thing that's kinda weird though...Kyosen-san, is it just me, or is that multiple-cpu result on Cinebench not correct? I recall getting a much higher number with my cpu's @ 2.0G. I'd run the bench again, but I think my Tyan s3993 has blown up (NO, it's not my fault :wasntme: ). Now I'm just sitting here waiting for some B2 Barcelonas and hopefully a new board.

Wonderful work, btw! :up:

EDIT: Ok, nevermind....you're only running one cpu....didn't notice that:p:

Just one more thing to add.....anyone notice the mobo box which his board is sitting on in picture one? ;) That's another reason why my cpu's are taking a "vacation" in Japan.

L1N64-SLI box eh? Sent that to him as well?

Quote:

---

Originally Posted by quake6

It is 7 second faster in super pi with just 400mhz up:eek:
Awsome :D

---

My first thougt too, but then again to get 9sek 1M you need something like a 11-1300MHz OC from 2GHz (3.1-3.3GHz)

Quote:

---

Originally Posted by SparkyJJO

L1N64-SLI box eh? Sent that to him as well?

---

Nope...kyosen doesn't need any help getting mobos :D ;)

All you guys comparing SuperPi scores seems forgetting that he is running Registered RAM! Only comparision to Opteron 2xxx with same memory will be valid. Still it is nice to know how relatively better Barcelona is compared to desktop K8...

Kyosen-san, could you please download this tool:

http://www.amd.com/us-en/assets/cont...r115_setup.zip

It will show us some info on the DPPM voltages and usage of the cpu. Do you have access to adjust NB voltage?

Quote:

---

Originally Posted by s7e9h3n

Kyosen-san, could you please download this tool:

http://www.amd.com/us-en/assets/cont...r115_setup.zip

It will show us some info on the DPPM voltages and usage of the cpu. Do you have access to adjust NB voltage?

---

My son,
didn't you have a RD790 board?:shrug:

Quote:

---

Originally Posted by BeardyMan

My son,
didn't you have a RD790 board?:shrug:

---

Hammerhead is an AM2+ platform, not socketF. I'm waiting for a Phenom to run on the board. I could run an AM2, but what's the point?

Quote:

---

Originally Posted by BeardyMan

My son,
didn't you have a RD790 board?:shrug:

---

His Phenoms are not there yet (at least that's what i think he said),so the board sits waiting for its master :D

Quote:

---

Originally Posted by informal

His Phenoms are not there yet (at least that's what i think he said),so the board sits waiting for its master :D

---

jinx :D

Quote:

---

Originally Posted by s7e9h3n

Hammerhead is an AM2+ platform, not socketF. I'm waiting for a Phenom to run on the board. I could run an AM2, but what's the point?

---

Are you expecting them soon or???:D

Quote:

---

Originally Posted by BeardyMan

Are you expecting them soon or???:D

---

Yes...very soon...along with B2 Barcelonas ;)

Quote:

---

Originally Posted by Lightman

All you guys comparing SuperPi scores seems forgetting that he is running Registered RAM! Only comparision to Opteron 2xxx with same memory will be valid. Still it is nice to know how relatively better Barcelona is compared to desktop K8...

---

Yes, I know. Thats why I said my comparison probably wasn't valid. But. considering all the disadvatages, K10, still appears quicker even when compaired to more advantageous memory systems.

If AMD built them to run SuperPie they would have named them Kobayashi. :D

32sec at 2.4GHz in SuperPi, that means -at the very best - 21sec at 3.6 GHz.
Much much slower than Core2 can do.
K10 is not going to be new SuperPi King for sure:)

Can the respected owner run some more SMP benches, like Frits Chess Benchmark, wprime, 9.1 Cinebench ?

AMD never been spi king. Even when Intel had netburst. But this is one of signs that K10 will be beast for sure.

Quote:

---

Originally Posted by Cronos

32sec at 2.4GHz in SuperPi, that means -at the very best - 21sec at 3.6 GHz.
Much much slower than Core2 can do.
K10 is not going to be new SuperPi King for sure:)

Can the respected owner run some more SMP benches, like Frits Chess Benchmark, wprime, 9.1 Cinebench ?

---

Hmmm how can you get 21s @ 3.6 ghz?

The memmory are still runing at very slow speed 401mhz 5-5-5
And 400mhz more core speed give 7.078s improvement in super pi.

2.0 ghz = 39.750
2.4ghz = 32,672 -7.078s
2.8ghz = 25,594 -7.078s (dont think this is correct)
3.2ghz = 18,516 -7.078s (dont think this is correct)
3.6ghz = 11,438 -7.078s (dont think this is correct)


This is what it will give if the scaling is linear, but i dont think it is. And Core2Duo will giving more boost in super pi1m becuse of larger L2.

We will still need to see how fast it is @ 2.8ghz in super pi 1m, and super p1m is not evry thing.

Nice chips :D

Cinebench score is better, although again not as good as on Core2. Single Core 2 Quad 3600MHz scores 14700 here.
It is more-or-less obvious that k10 is not going to reach Core2 performance level per clock. Superior memory subsystem can give it significant advantage on server platform, but the desktop is still ruled by Intel Core2. Lets hope AMD will be able to overcome GHz shortage and compete here.

Quote:

---

Originally Posted by Lastviking

Hmmm how can you get 21s @ 3.6 ghz?

The memmory are still runing at very slow speed 401mhz 5-5-5
And 400mhz more core speed give 7.078s improvement in super pi.

2.0 ghz = 39.750
2.4ghz = 32,672 -7.078s
2.8ghz = 25,594 -7.078s (dont think this is correct)
3.2ghz = 18,516 -7.078s (dont think this is correct)
3.6ghz = 11,438 -7.078s (dont think this is correct)

---

Too bad you missed your math classes in the school. Sometimes, you can learn something really usefull there :)
50% increase in CPU clock speed (2.4->3.6) gives you 50% improvement in execution times. That is, provided the memory is speeded up accordingly, which is not the case here.
32/1.5=21.3, thats the best we can expect from K10 at 3.6 GHz. Faster memory can win, maybe, 1-2 seconds here -no more.

Quote:

---

Originally Posted by Lastviking

Hmmm how can you get 21s @ 3.6 ghz?

The memmory are still runing at very slow speed 401mhz 5-5-5
And 400mhz more core speed give 7.078s improvement in super pi.

2.0 ghz = 39.750
2.4ghz = 32,672 -7.078s
2.8ghz = 25,594 -7.078s (dont think this is correct)
3.2ghz = 18,516 -7.078s (dont think this is correct)
3.6ghz = 11,438 -7.078s (dont think this is correct)


This is what it will give if the scaling is linear, but i dont think it is. And Core2Duo will giving more boost in super pi1m becuse of larger L2.

We will still need to see how fast it is @ 2.8ghz in super pi 1m, and super p1m is not evry thing.

---

And at 4.4Ghz we can expect -2.718s 1M Super Pi! Phenom can travel through time :) before you click you have the answer!

Quote:

---

Originally Posted by Cronos

Too bad you missed your math classes in the school. Sometimes, you can learn something really usefull there :)
50% increase in CPU clock speed (2.4->3.6) gives you 50% improvement in execution times. That is, provided the memory is speeded up accordingly, which is not the case here.
32/1.5=21.3, thats the best we can expect from K10 at 3.6 GHz. Faster memory can win, maybe, 1-2 seconds here -no more.

---

Is the scaling same on all cpus?K8, Core2Duo, K10? . And like i wrote before you see the ( ) next time you dont skip your read classes :)

We need more tests before we can make how much it will scale, and if all stuff works likes it should be working.

Writing 21,3s is wrong before we know.

2.0ghz = 39.750
2.4ghz = 32,672 ~7.0s
2.8ghz = 27,672 ~5.0s (dont think this is correct)
3.2ghz = 23,672 ~4.0s (dont think this is correct)
3.6ghz = 20,672 ~3.0s (dont think this is correct)

Correcting abit...after some testing, but i think this is fair away from right. The show before was if evry 400mhz of K10 give 7s like someone write before thats not are correct(like Andi and one more before wrote it will give -times to early ^^

Quote:

---

Originally Posted by Cronos

Cinebench score is better, although again not as good as on Core2. Single Core 2 Quad 3600MHz scores 14700 here.
It is more-or-less obvious that k10 is not going to reach Core2 performance level per clock. Superior memory subsystem can give it significant advantage on server platform, but the desktop is still ruled by Intel Core2. Lets hope AMD will be able to overcome GHz shortage and compete here.

---

Thats strange I just looked at the Anand article on the G0 stepping q6600 and at 3.33G it scored 1792 for cinebench. That was on cinebench 9.5, but I can't see the score jumping that much on version 10. I would bet the 14700 is the muticore score and we are talking about 2.4G vs 3.6G. What does the q6600 score stock is the question.

Quote:

---

Originally Posted by PhilDoc

Thats strange I just looked at the Anand article on the G0 stepping q6600 and at 3.33G it scored 1792 for cinebench. That was on cinebench 9.5, but I can't see the score jumping that much on version 10. I would bet the 14700 is the muticore score and we are talking about 2.4G vs 3.6G. What does the q6600 score stock is the question.

---

Q6600 G0 @3600MHz
http://img81.imageshack.us/img81/907...4q36co8.th.png

At 3.6GHz, the K10 will score no more than 7709*1.5=11563

Quote:

---

Originally Posted by PhilDoc

Thats strange I just looked at the Anand article on the G0 stepping q6600 and at 3.33G it scored 1792 for cinebench. That was on cinebench 9.5, but I can't see the score jumping that much on version 10. I would bet the 14700 is the muticore score and we are talking about 2.4G vs 3.6G. What does the q6600 score stock is the question.

---

Cinebench R10 is significantly different. The stock Q6600 in 32-bit Cinebench R10 scores a bit higher than 2400/8600.

Agreed from looking at the q6600 scores cinebench 10 is different, but the single core score for the q6600 at 3.6G is 4471 not 14700. Just as I thought that was the muticore score. Certainly still quicker, but I'd still like to see its score at stock.

Quote:

---

Originally Posted by s7e9h3n

Yes...very soon...along with B2 Barcelonas ;)

---

How come no one shouted out OMGZ !!! when steven said B2 Barcelonas...are people THAT disconnected when it comes to amd 0_o...

B2 Barcelonas = penryn fighting power people.

Perkam

Quote:

---

Originally Posted by Lastviking

Is the scaling same on all cpus?K8, Core2Duo, K10? . And like i wrote before you see the ( ) next time you dont skip your read classes :)

---

The scaling is different of course, but can differ only to the lower numbers, not higher. I.e.
50% more clock speed can give you anywhere from 0 to 50% improvement in scores/exec times.

Thanks kyosen-san. :)

Here's the crack so far.

Tech Report review showed Barcelona scaling more than 100% in performance in comparison to core speed in a few cases. Because it's not just core speed we're talking about, but L3 cache speed/latency and IMC speed.

SPI/Cinebench 10 (incl. S7s old 8-core CB tests)

2GHz K10 39.750s - 2.4GHz K10 32.672s
(2GHz) 8 cores/DDR2-667/NB1.6GHz CB 1907/12941 - 4 cores/DDR2-800/NB1.8GHz CB 1984/7709

SPI Scaling
20% clock change ((2400-2000)/2000 x 100)
17.8% performance increase ((39.750-32.672)/39.750 x 100)

17.8/20 x 100 = 89% scaling

Quote:

---

Originally Posted by KTE

SPI Scaling
20% clock change ((2400-2000)/2000 x 100)
17.8% performance increase ((39.750-32.672)/39.750 x 100)

17.8/20 x 100 = 89% scaling

---

Actually it should be ((39.750-32.672)/32.675 = 21.7%.

The speedup is higher than 20%, but this is just withing error margin.
There is one reason there speedup may be higher than increase in clock speed -if there are some background application(s) clogging the cpu.
Higher CPU clock will leave more resources for the main application.

Quote:

---

Originally Posted by Cronos

Actually it should be ((39.750-32.672)/32.675 = 21.7%.

The speedup is higher than 20%, but this is just withing error margin.
There is one reason there speedup may be higher than increase in clock speed -if there are some background application(s) clogging the cpu.
Higher CPU clock will leave more resources for the main application.

---

KTE's equation is correct. you use the higher score on the bottom.

The score I saw was 39.750 and I just looked again. His math is fine, I get the same 17.8% increase in spi time.

Quote:

---

Originally Posted by PhilDoc

The score I saw was 39.750 and I just looked again. His math is fine, I get the same 17.8% increase in spi time.

---

For dual-channel mode Spi 1M time was 38.968, and it is clear that overclocked results were taken with memory in dual-channel mode.
So, 38.968 / 32.672 = 1.1927 = 19.3% it is.

Quote:

---

Originally Posted by Cronos

For dual-channel mode Spi 1M time was 38.968, and it is clear that overclocked results were taken with memory in dual-channel mode.
So, 38.968 / 32.672 = 1.1927 = 19.3% it is.

---

can't you see the Single word written on cpu-z at the ram tab?

Quote:

---

Originally Posted by perkam

How come no one shouted out OMGZ !!! when steven said B2 Barcelonas...are people THAT disconnected when it comes to amd 0_o...

B2 Barcelonas = penryn fighting power people.

Perkam

---

I'm just not getting my hopes up, I'd love B2 to be much better, but if I don't get my hopes up then there isn't the chance of me crashing to the ground in case my hopes were in vain.

Quote:

---

Originally Posted by mrcmtl

can't you see the Single word written on cpu-z at the ram tab?

---

It was set to dual channel in the bios, CPU-Z seems to be reporting it wrong (as it is the HT link as well).

Quote:

---

I'd love B2 to be much better

---

Then let me tell you in advance, it will :D

Perkam

Quote:

---

Originally Posted by quake6

It is 7 second faster in super pi with just 400mhz up:eek:
Awsome :D

---

I noticed that too. Quite a good showing for a small clock speed increase.

Quote:

---

Originally Posted by Cronos

For dual-channel mode Spi 1M time was 38.968, and it is clear that overclocked results were taken with memory in dual-channel mode.
So, 38.968 / 32.672 = 1.1927 = 19.3% it is.

---

Yep, I finally saw it. I was just looking at the screen shots. Not totally sure he's actually running dual channel as cpu-z show single. It could be that cpu-z needs an upgrade for Barcelona. The difference in times may be just fluctuation in the spi test. We'll find out soon I suppose.

3.88 speeup on cinebench.. is it good enough compare to c2q?

Quote:

---

Originally Posted by Cronos

Actually it should be ((39.750-32.672)/32.675 = 21.7%.

---

That's 108% scaling. Its incorrect because you're not working out percentages here, but percentage change. The formula is as follows

Maths:
http://www.gcseguide.co.uk/percentages.htm
http://mathforum.org/library/drmath/view/63055.html

New - Old / Old x 100 = Percentage Change

Quote:

---

Originally Posted by MR_SmartAss

KTE, your math is wrong. Also, the SuperPi 1M time is 38.968s for 2GHz. Read the first post.

---

For single channel it's 39.750s for dual channel its what you said: http://www.xtremesystems.org/forums/...22&postcount=1

For simplicity, we swap the top figures around or it would become a minus change this time because of importance is decreasing values, not increasing (which is practically untrue i.e. to us 7 sec dropped is not 7 disadvantage but 7 advantage). The base always stays the same as its "new" relative to "old". What matters is the change.

38.968s - 32.672s (change) / 38.968s (relative to old) x 100 (percentage) = 16.157%

If you use that first general formula it would be -16.157% change which is perfectly true from the original time. But to us that's a +16.157% performance gain.

Now we can use a simple percentage formula:

16.157 / 20 x 100 = 80.8% scaling.

Quote:

---

Originally Posted by tictac

3.88 speeup on cinebench.. is it good enough compare to c2q?

---

It is unreachable for С2Q. The main reason is Core2 shared cache.
This is also one of the main reasons for excellent single-threaded performance of Core2 cpu's, which -in turn - is unreachable for comparatively clocked K10, regardless of stepping.
The net result is that C2Q is excellent CPU with very high SMP performance. It's
"untrue" nature does not hurt its performance to any noticeable extent. No reason for Intel to hurry up with "native" solution -at least, not from performance stand point.

Quote:

---

Originally Posted by s7e9h3n

Hammerhead is an AM2+ platform, not socketF. I'm waiting for a Phenom to run on the board. I could run an AM2, but what's the point?

---

Yep, you actually want a Wahoo board for dual f1207 sockets :)

FELIX kindly pointed out to me about K10's memory controller:)
According to AMD's document, BIOS and Kernel　Developer’s Guide,
K10's has 2 DCTs(DCT: Dram ConTroller!?), and it can work different two mode, i.e.,
(1) to behave as a single dual-channel DCT; this is called ganged mode;
(2) to behave as two single-channel DCTs; this is called unganged mode.

My understanding is:
Ganged mode reads or writes with 2 DCTs at the same time.
Unganged mode also can reads with 1 DCT and writes with another 1 DCT at the same time,
if the system requests as so.
Unganged mode is more flexible than ganged mode,
and ganged mode is equal to "conventional" dual channel mode.

Then, I noticed the BIOS setting, "Unganged Mode support".
http://222.151.153.254/c-board/file/...S_unganged.jpg

So I tried SuperPI 4M in 3 case as below, and results:
with 1 memory module: 3m27.141s
with 2 memory modules, ganged mode: 3m24.438s
with 2 memory modules, unganged mode: 3m22.953s
and we can see CPU-Z and FELIX's test program properly detects
"conventional" dual channel mode:up:
Default BIOS setting is "Unganged Mode = Enabled", and at least,
unganged mode looks faster than ganged mode for SuperPI.
http://222.151.153.254/c-board/file/...m_unganged.png
http://222.151.153.254/c-board/file/...mem_ganged.png
http://222.151.153.254/c-board/file/...m_unganged.png

And these are latecy test results with Everest Ultimate v4.20:
at 2.0G=200x10: 76.0ns
at 2.2G=220x10: 69.0ns
at 2.4G=240x10: 63.4ns
The latest Everest detects Barcelona properly:)
http://222.151.153.254/c-board/file/...10_latency.png
http://222.151.153.254/c-board/file/...10_latency.png
http://222.151.153.254/c-board/file/...10_latency.png

Thanks to everyone for many comments and suggestions:)
...but don't be bullying poor small AMD, hipro5:D (yeah, just kidding;) )

Quote:

---

Originally Posted by KTE

38.968s - 32.672s (change) / 38.968s (relative to old) x 100 (percentage) = 16.157%

---

What if the faster score is 20s? With your equation, it's only a 49% increase in performance.

Or look at it this way, the 2GHz does 1M digits in 38.968s = 25,662 digits/s; the 2.4GHz does 1M digits in 32.672s = 30607 digit/s. The 2.4GHz does 19.2% more digits/s.

Please calm down about scaling for SuperPI time...
I'll try it again, with K8 Opteron 2212 and K10, with same OS:yepp:

We are all calm. Somebody just needs to learn their math lessons, thats all :)

It is much more interesting to see multi-threaded scores of dual-cpu K10 system,
as it is there AMD is going to compete and win against Intel -not on single-socket platform.

ny chance of a pifast bench? That was always more favourable to AMD than SuperPi and also seemed a bit more relevant to general useage.

Regards

Andy

Quote:

---

Originally Posted by quake6

It is 7 second faster in super pi with just 400mhz up:eek:
Awsome :D

---

Actually, it is about 20% faster for a 20% faster clock... this is expected... if it didn't something is very wrong.

Wow, not bad actually. I get 32sec @ 2.9ghz on opty 165. So that means a 3ghz barc. should do around 25 sec, not bad at all.

Interesting though, how a 6000+ A64 is par with a C2d 6600 model in real time performance but in SuperPi the 6600 smokes it. But in this case a 2.4ghz Phenom should smoke the hell out of the 6600 out the water in real time applications.

Quote:

---

Originally Posted by JumpingJack

Actually, it is about 20% faster for a 20% faster clock... this is expected... if it didn't something is very wrong.

---

It depends how you benchmark each application, most of these here are singel threaded. :up:

how about kribibench?

http://www.xtremesystems.org/forums/...89#post2476489

Great results kyosen. Keep 'em coming. :)

Quote:

---

Originally Posted by KTE

Great results kyosen. Keep 'em coming. :)

---

Agreed, now we want higher clocks :D
Higher clocks = Answer to the important questions
;)

can you do same comparison on the memory mode with 64bit cinebench

These are quite promising, Thanks aswell Kyosen

Quote:

---

Originally Posted by kyosen

Thanks to everyone for many comments and suggestions:)
...but don't be bullying poor small AMD, hipro5:D (yeah, just kidding;) )

---

I won't for sure.....:D :p: