Discussion X5470 to X5472

Will performance increase?

  • Yes, performance will increase.

    Votes: 0 0.0%
  • No, performance will be about the same.

    Votes: 0 0.0%
  • No, performance will decrease.

    Votes: 0 0.0%

  • Total voters
    0

prince_xaine

Reputable
Hello everyone! I am going to be "upgrading" a system running a Xeon X5470 to a Xeon X5472 and figured this would make for a neat discussion. Do you think there will be any performance increases/decreases and why?
Note: This is obviously a project for fun and I have a modern system. The motherboard and processors are from the 2006-2008 era.

The current specs are:
OS: Windows 11 64-Bit
Mainboard: Asus P5Q3-Deluxe
Processor: Intel X5470 =>Intel X5472
Memory: G. Skill F3-12800CL9D-16GBXL (4x4)
Graphics: Zotac Amp! Nvidia GTX 1080 (10DE 1B80)
Storage: Samsung 860 Evo 250 GB
Storage: WD WD6003FZBX-00K5WB0 6TB (RAID 0)
Storage: WD WD6003FZBX-00K5WB0 6TB (RAID 0)
Power: EVGA SuperNOVA 850 P2
Cooling: Noctua NH-D14
Userbenchmark: https://www.userbenchmark.com/UserRun/52368909

More about the system:
The system is running Windows 11 64-Bit on a Samsung 860 Evo 250 GB SSD. The SSD is limited by the SATA II interface. The games will be running on the 12 TB RAID 0 HDDs, which sequentially can peak over 400 MB/s and run sustained speeds of around 300 MB/s, which slightly outperforms the SATA II interface. The mainboard FSB is set at 448 MHz, which also overclocks both the processor and memory. The X5470, with a FSB ratio of 10:1, is running at a current speed of 4.48 GHz, and the memory is running at 1494 MHz. The processor at 4.48 GHz is mostly stable, and performs much like the i3-8100. The graphics card is limited by the PCIe 2.0 x16 interface, which causes some performance drops in bandwidth hungry titles, and is especially noticeable in DX9 and DX12 applications. Most titles (except those that are especially CPU intensive) run at around 60-80 FPS, but can be much higher (like DOOM 2016 - which runs between 150 FPS - 200 FPS) with more optimized titles.

Things to consider:
  1. The X5470's supported bus speed is 1333 MHz, whereas the X5472's supported bus speed is 1600 MHz.
  2. The X5470 (with a ratio of 10:1) runs at a base clock speed of 3.33 GHz, the X5472 (with a ratio of 7.5:1) runs at a base clock speed of 3.00 GHz.
  3. The current X5470 is a high bin chip, and runs about 0.30 GHz higher than most X5470 chips.
  4. The BIOS microcode has been manually updated for the processors, and the LGA 771 processors have been modified to fit and run in the LGA 775 socket.
Test Environment:
Benchmarks will be run following the following standards:

  1. First wave of benchmarks and game testing will place both processors at the maximum capable speed of the X5472. I will have a few of the processors, and will use the processor that overclocks the highest.
  2. Second wave of benchmarks and game testing will compare the maximum capable speed of the X5470 to the X5472. I entirely expect the X5470 to have a higher clock speed.
  3. Benchmarks being used will be: Userbenchmark, 3DMark, Heaven, Passmark, Cinebench, CPU-Z. Games are TBD.
I am willing to accept suggestions for game benchmarks. Keep in mind that certain titles may not run due to the lack of instructions on the outdated processors.
 
Last edited:

prince_xaine

Reputable
While the difference is largely negligible, the X5470 is marginally faster overall.

https://cpu.userbenchmark.com/Compare/Intel-Xeon-X5472-vs-Intel-Xeon-X5470/m3914vsm14102
The performance comparison on userbenchmark is assuming the processors will be running at their base clock speeds. I will be running 2 waves of tests, the first wave will match the processors to the same clock speed. The second wave will allows the X5470 to run at 4.48 GHz. While the X5470 will most likely triumph in the second wave, do you think the X5472 will outperform the X5470 at the same clock speed?
 

prince_xaine

Reputable
The interesting aspect of this is that the processor clock speed difference is about 11%, and the processor single thread performance difference is 13.4%, which is well within the margin of error. It's interesting to find out if increasing the processor bus bandwidth from 1333 MHz to 1600 MHz will allow other performance increases outside of processor performance alone.
 

USAFRet

Titan
Moderator
Mar 16, 2013
157,529
12,076
176,090
24,359
The interesting aspect of this is that the processor clock speed difference is about 11%, and the processor single thread performance difference is 13.4%, which is well within the margin of error. It's interesting to find out if increasing the processor bus bandwidth from 1333 MHz to 1600 MHz will allow other performance increases outside of processor performance alone.
For me, that amount of difference is not worth the hour or two taking it apart to change the CPU.
 

prince_xaine

Reputable
If you think it is worth it, got for it.

The only one you have to please is you.
Fair enough. Building computers, making small tweaks to get just a little bit more performance and benchmarking are all fun hobbies for me. Not everyone feels the same way about that, and I can respect it. Was just explaining how quick the process would be, I don't think I'd do this if it took an hour or two just to change the processor. Now it will take an hour or two of tweaking and benchmarks to come up with results., but I find that part quite intriguing.
 

ASK THE COMMUNITY