Intel NUC has much worse performance than identical system

Question

I have two Intel NUC systems:

2.4 GHz i3-7100U (2.40 GHz - 3MB Cache - 2 Cores - 4 Threads)
4 GB DDR4 at 2133 MHz (1× 4 GB)
250 GB NVMe PCIe M.2 SSD

I bought the first one last year and it is performing quite well. I've had the second for a few days now and performance is much worse.

To eliminate all variables for testing, I've booted up both systems with the same live flash drive (Pop!_OS), and run Minetest. The older system gets 60 fps while the new one gets only 20. As far as I can tell, bios settings are the same.

What could cause such a difference in graphics performance?

Update - I've added information from dmidecode

From the old system (purchased last year):

Handle 0x0000, DMI type 0, 24 bytes
BIOS Information
    Vendor: Intel Corp.
    Version: BNKBL357.86A.0057.2017.1122.1550
    Release Date: 11/22/2017
    Address: 0xF0000
    Runtime Size: 64 kB
    ROM Size: 8192 kB
    Characteristics:
        PCI is supported
        BIOS is upgradeable
        BIOS shadowing is allowed
        Boot from CD is supported
        Selectable boot is supported
        BIOS ROM is socketed
        EDD is supported
        5.25"/1.2 MB floppy services are supported (int 13h)
        3.5"/720 kB floppy services are supported (int 13h)
        3.5"/2.88 MB floppy services are supported (int 13h)
        Print screen service is supported (int 5h)
        Serial services are supported (int 14h)
        Printer services are supported (int 17h)
        ACPI is supported
        USB legacy is supported
        BIOS boot specification is supported
        Targeted content distribution is supported
        UEFI is supported
    BIOS Revision: 5.6
    Firmware Revision: 8.12

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
    Manufacturer: Intel Corporation
    Product Name: NUC7i3BNB
    Version: J22859-308
    Serial Number: GEBN730007PT
    Asset Tag:                                  
    Features:
        Board is a hosting board
        Board is replaceable
    Location In Chassis: Default string
    Chassis Handle: 0x0003
    Type: Motherboard
    Contained Object Handles: 0

From the new system:

Handle 0x0000, DMI type 0, 24 bytes
BIOS Information
    Vendor: Intel Corp.
    Version: BNKBL357.86A.0057.2017.1122.1550
    Release Date: 11/22/2017
    Address: 0xF0000
    Runtime Size: 64 kB
    ROM Size: 8192 kB
    Characteristics:
        PCI is supported
        BIOS is upgradeable
        BIOS shadowing is allowed
        Boot from CD is supported
        Selectable boot is supported
        BIOS ROM is socketed
        EDD is supported
        5.25"/1.2 MB floppy services are supported (int 13h)
        3.5"/720 kB floppy services are supported (int 13h)
        3.5"/2.88 MB floppy services are supported (int 13h)
        Print screen service is supported (int 5h)
        Serial services are supported (int 14h)
        Printer services are supported (int 17h)
        ACPI is supported
        USB legacy is supported
        BIOS boot specification is supported
        Targeted content distribution is supported
        UEFI is supported
    BIOS Revision: 5.6
    Firmware Revision: 8.12

Handle 0x0002, DMI type 2, 15 bytes
Base Board Information
    Manufacturer: Intel Corporation
    Product Name: NUC7i3BNB
    Version: J22859-312
    Serial Number: GEBN809003J5
    Asset Tag:                                  
    Features:
        Board is a hosting board
        Board is replaceable
    Location In Chassis: Default string
    Chassis Handle: 0x0003
    Type: Motherboard
    Contained Object Handles: 0

Update 2

I ran a simple ram speed test found at https://www.pcsuggest.com/benchmarking-linux-systems-with-simple-command-line-tools/

the script:

mkdir ramtest
sudo mount tmpfs -t tmpfs ramtest/
cd ramtest
dd if=/dev/zero of=data_tmp bs=1M count=512
dd if=data_tmp of=/dev/null bs=1M count=512
cd ..
sudo umount ramtest

results on the old system:

512+0 records in
512+0 records out
536870912 bytes (537 MB, 512 MiB) copied, 0.225257 s, 2.4 GB/s
512+0 records in
512+0 records out
536870912 bytes (537 MB, 512 MiB) copied, 0.0939961 s, 5.7 GB/s

results on the new system:

512+0 records in
512+0 records out
536870912 bytes (537 MB, 512 MiB) copied, 1.28991 s, 416 MB/s
512+0 records in
512+0 records out
536870912 bytes (537 MB, 512 MiB) copied, 0.602888 s, 890 MB/s

Are both exactly the same hardware including the board revision *and* same UEFI version (what you call BIOS but really isn't)? — , Sep 01 '18 at 00:09
Do you have a model number or something to reference the installed hardware. — Hogstrom, Sep 01 '18 at 00:31
I've added information to the question. Version numbers aren't identical and I'm hesitant to upgrade the UEFI on a system that is working well right now. — Jon Hulka, Sep 01 '18 at 03:04
I don't know if it might have to do with spectre/meltdown mitigations, but also compare `uname -a` and `grep 'model\|stepping\|microcode\|bugs' /proc/cpuinfo`. — aventurin, Sep 01 '18 at 18:33
@aventurin both systems give identical output on those commands, but your comment prompted me to find the answer. My computer's distributor had installed firmware version 57. The latest version is 67. I think 57 came out just after the vulnerabilities were discovered and was a bit rushed, so not performance tuned. I installed 67 from Intel's site and it is screaming fast now. — Jon Hulka, Sep 01 '18 at 23:09
@aventurin since your comment prompted the solution, if you write it up as an answer I will mark it as the correct one. — Jon Hulka, Sep 01 '18 at 23:10
Just FYI, I also had graphics performance issues, though likely a different cause to yours. Solution eventuated to be removing the cooling fan and finding dust trapped where it connects to the cooling fins. https://superuser.com/questions/1363932/intel-nuc-overheating-hot-video-stuttering/1363933#1363933 — Mtl Dev, Oct 04 '18 at 17:12

score 1 · Accepted Answer · answered Sep 02 '18 at 00:14

1

The performance problem is, as you confirmed, related to the old version 57 firmware.

Installing the newest firmware version 67 solves the problem.

This might be due to improved microcode updates after the January Meltdown/Spectre mitigation updates.

answered Sep 02 '18 at 00:14

aventurin

236
1
3
8

1

Thanks for your help on this. I've been tearing my hair out for three days now. I have no idea how many more days of anguish you saved me. – Jon Hulka Sep 02 '18 at 00:34

Intel NUC has much worse performance than identical system

1 Answers1