Tweet
Gentlemen?,
Before upgrading a Dell Precision T5500 that I bought to replace a T5400 second system, I ran PT on the T5500 in it's original Xeon E5620 (4-core @ 2.4 /2.6GHz) configuration. The modest system rating of 1479 was not too big a concern as I knew the Quadro FX 580 would have quite a low 3D score and given the weighting of the 3D benchmark in the system score.
What did concern me was that the CPU score was 4097. The Passmark average for the E5620 is 4929.
Doing an advanced baseline search: "Other: Model:=T5500 / "CPU = E5620", there were 40 results with 13 single E5620 systems. The single E5620 systems have a high CPU score of 5245 and low of 3645. That's a wide range- 44%- for the same CPU on the same motherboard. I didn't think differences in GPU, RAM, and disk could account for the variation.
I went ahead and changed the CPU to a Xeon X5680 (6-core @ 3.33 / 3.6), changed from 6GB to 24GB ECC 1333, to the spare Quadro 4000 and had a similarly disappointing rating of 1797 and CPU score of 7612 The Passmark average for the E5620 is 9275. Of the 7 baseline T5500 with the X5680, 3 had single CPU's, scoring: 9226, 7537, and 7519. The ratings were not too dramatically different: 2471, 2254, and 2460- the system with the lowest CPU score (7519) had similar rating (2460) to the system with the 9226 CPU and 2471 rating. Odd! The system with CPU= 9226 had a disk score of 2189 and the other two were 1206 and 1115 but GPUs were tighter: the lowest CPU score had a 916 in 3D and the 9226 CPU system had 689 3D. This suggested the low score was centered to some feature or setting of the CPU. Yes, this made me dizzy too!
Why was an older but still quite capable CPU doing this? I thought of power settings but I'd already set the T5500 to High Performance in Power Options. I had also upgraded BIOS from AO6 to A16.
Then, it occurred to me that the other two low scoring T5680 systems may have also been upgraded and a setting had been carried over from the original. And there was the answer in BIOS > enable hyperthreading. I reran PT and there it was, a CPU of 9327.
Changing the RAM to 24GB ECC 1333 and adding the Quadro 4000 produced a new system rating of 2932 and I expect that to improve again when I figure out how to add the Samsung 840 to the PERC 6/i SAS controller- I think this means an SAS to SATA interposer on the HDD0 SAS cable/ connector to correct the signal rate, but if anyone reading knows, it would be a help. > My first SAS drive and SAS/SATA controller.
Sorry for the long ramble, but I thought to relate another success story- and big relief- based on looking closely into Performance Test results. I've seen a lot of the TX500 series Precisions in the baselines with noticeably lower CPU scores and it appears there may be a 20-40% CPU boost with the one magic setting.
PT is incredibly useful- a fantastic shortcut to decision-making for those of us with limited experience.
Cheers,
BambiBoom
Before upgrading a Dell Precision T5500 that I bought to replace a T5400 second system, I ran PT on the T5500 in it's original Xeon E5620 (4-core @ 2.4 /2.6GHz) configuration. The modest system rating of 1479 was not too big a concern as I knew the Quadro FX 580 would have quite a low 3D score and given the weighting of the 3D benchmark in the system score.
What did concern me was that the CPU score was 4097. The Passmark average for the E5620 is 4929.
Doing an advanced baseline search: "Other: Model:=T5500 / "CPU = E5620", there were 40 results with 13 single E5620 systems. The single E5620 systems have a high CPU score of 5245 and low of 3645. That's a wide range- 44%- for the same CPU on the same motherboard. I didn't think differences in GPU, RAM, and disk could account for the variation.
I went ahead and changed the CPU to a Xeon X5680 (6-core @ 3.33 / 3.6), changed from 6GB to 24GB ECC 1333, to the spare Quadro 4000 and had a similarly disappointing rating of 1797 and CPU score of 7612 The Passmark average for the E5620 is 9275. Of the 7 baseline T5500 with the X5680, 3 had single CPU's, scoring: 9226, 7537, and 7519. The ratings were not too dramatically different: 2471, 2254, and 2460- the system with the lowest CPU score (7519) had similar rating (2460) to the system with the 9226 CPU and 2471 rating. Odd! The system with CPU= 9226 had a disk score of 2189 and the other two were 1206 and 1115 but GPUs were tighter: the lowest CPU score had a 916 in 3D and the 9226 CPU system had 689 3D. This suggested the low score was centered to some feature or setting of the CPU. Yes, this made me dizzy too!
Why was an older but still quite capable CPU doing this? I thought of power settings but I'd already set the T5500 to High Performance in Power Options. I had also upgraded BIOS from AO6 to A16.
Then, it occurred to me that the other two low scoring T5680 systems may have also been upgraded and a setting had been carried over from the original. And there was the answer in BIOS > enable hyperthreading. I reran PT and there it was, a CPU of 9327.
Changing the RAM to 24GB ECC 1333 and adding the Quadro 4000 produced a new system rating of 2932 and I expect that to improve again when I figure out how to add the Samsung 840 to the PERC 6/i SAS controller- I think this means an SAS to SATA interposer on the HDD0 SAS cable/ connector to correct the signal rate, but if anyone reading knows, it would be a help. > My first SAS drive and SAS/SATA controller.
Sorry for the long ramble, but I thought to relate another success story- and big relief- based on looking closely into Performance Test results. I've seen a lot of the TX500 series Precisions in the baselines with noticeably lower CPU scores and it appears there may be a 20-40% CPU boost with the one magic setting.
PT is incredibly useful- a fantastic shortcut to decision-making for those of us with limited experience.
Cheers,
BambiBoom
Comment