Gen 10 Intel CPU seems slow?

Discussion in 'Troubleshooting Help' started by munted123, Apr 27, 2021.

  1. munted123

    munted123 Member

    Joined:
    Aug 21, 2001
    Messages:
    61
    Hey all, I don't know much about recent Intel CPUs but I just got a i3-10105F to replace my old i7 3770, all the benchmarks I saw seemed to indicate it'd be similar performance to an i7 7700. But when I did my own gaming benchmarks my new i3-10105F seems slower than my old i7 3770 overall.

    Uniengine Superposition less than 1% difference
    i7 3770 and R9 290x: Score: 8174, FPS Min: 51.47, Ave: 61.14, Max: 76.57. GPU Min 29 deg, Max 69 deg. GPU Utilization: 100%
    i3 10105F and R9 290x: Score: 8133, FPS Min: 50.46, Ave: 60.84, Max: 76.48. GPU Min 30.0 deg, Max 70.0 deg. GPU Utilization: 100%

    3dMark Time Spy, 1.5% faster
    i7 3770 and R9 290x: Score 3958. Graphics score 3998. CPU score 3748.
    i3 10105F and R9 290x: Score 4014. Graphics score 3985. CPU score 4192.

    Company of Heroes 2, 3% slower
    i7 3770 and R9 290x: Min 42.57, Max 103.28, Ave 65.23
    i3 10105F and R9 290x: Min 41.30, Max 93.81, Ave 58.37

    Ashes of the Singularity Escalation, 17% slower
    i7 3770 and R9 290x: Ave Framerate: 56.2. Ave CPU Framerate: 59.7 FPS.
    i3 10105F and R9 290x: Ave Framerate: 47.0. Ave CPU Framerate: 47.1 FPS

    *Assetto Corsa 6% faster
    i7 3770 and R9 290x: 145 Ave FPS, 69 Min, 189 Max, Variance=4 CPU=72%
    i3 10105F and R9 290x: Ave 154 FPS, 50 Min, 203 Max, Variance=15 CPU=89%

    *Assassin's Creedy Odyssey 12% slower
    i7 3770 and R9 290x: FPS: 46, Min 21, Max 75
    i3 10105F and R9 290x: FPS: 38, Min 22, Max 54

    I did a brand new Win10 x64 install and the latest drivers from the MSI website for the motherboard and the AMD website for my R9 290X GPU.

    Old system specs are:
    i7 3770 CPU
    R9 290x GPU
    Asrock Z68 motherboard
    16GB DDR3 1600MHz

    New system specs are:
    i3 10105F
    R9 290x
    MSI B460 motherboard
    16GB DDR4 2666MHz - motherboard won't allow RAM to run faster

    Looking at the CPUs compared: https://ark.intel.com/content/www/us/en/ark/compare.html?productIds=65719,203474
    The new CPU is also 4 cores, 8 threads, 3.7GHz vs 3.4GHz base, 4.4GHz vs 3.9GHz turbo, 14nm vs 22nm and 7 generations newer. The only drawback is 6MB cache vs 8MB cache.

    Can anyone think of anything I should look for?
     
  2. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    11,421
    The Gpu is the limitation.
     
  3. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    11,421
    By that, I mean run some Cpu tests. I doubt the CPU is slower in practice. There might be some unnoticed settings that auto-gimp on the old platform but not the new.
     
  4. Groff

    Groff Member

    Joined:
    Jun 22, 2012
    Messages:
    284
    Location:
    NSW
    Interesting as I have a 3770 as my main office machine and have resisted to upgrade that box.
     
  5. OP
    OP
    munted123

    munted123 Member

    Joined:
    Aug 21, 2001
    Messages:
    61
    Thanks for the replies guys, so there's a setting in the BIOS where you can set the CPU cooler type; stock, tower air or water cooler. From memory it changes the power limit to 100amp, 160amp or 160amp depending on what you select. There wasn't much difference in the max clock speed before I changed it 4200MHz vs 4300MHz but I assume it says at the max turbo for longer. Anyway I've put the new benchmarks below PL+ is now that I've changed the cooler type in BIOS.

    But in summary the i3 10105F has gone from 4% slower overall across 6 benchmarks to 2% faster now, you can see below there's a noticeable performance difference in several benchmarks. I suppose I could put my 5700XT GPU in and see how much difference a faster GPU would make but I guess I'm ok with the results for now. When it was slower I knew there was something wrong, even if the 290X is a bottleneck you shouldn't get worse benchmarks, not much improvement sure but slower, nah.

    Uniengine Superposition less than 1% slower before, 2% faster after changing PL

    i7 3770 and R9 290x: Score: 8174, FPS Min: 51.47, Ave: 61.14, Max: 76.57. GPU Min 29 deg, Max 69 deg. GPU Utilization: 100%
    i3 10105F and R9 290x: Score: 8133, FPS Min: 50.46, Ave: 60.84, Max: 76.48. GPU Min 30.0 deg, Max 70.0 deg. GPU Utilization: 100%
    PL+ i3 10105F and R9 290x: Score: 8339, FPS Min: 52.28, Ave: 62.37, Max: 77.06. GPU Min 35.0 deg, Max 74.0 deg. GPU Utilization: 100%

    3dMark Time Spy, 1.5% faster before, 3% faster after
    i7 3770 and R9 290x: Score 3958. Graphics score 3998. CPU score 3748.
    i3 10105F and R9 290x: Score 4014. Graphics score 3985. CPU score 4192.
    PL+ i3 10105F and R9 290x: Score 4091. Graphics score 4049. CPU score 4353.

    Company of Heroes 2, 3% slower before, 2% faster after
    i7 3770 and R9 290x: Min 42.57, Max 103.28, Ave 65.23
    i3 10105F and R9 290x: Min 41.30, Max 93.81, Ave 58.37
    PL+ i3 10105F and R9 290x: Min 43.61, Max 108.46, Ave 69.06ha

    Ashes of the Singularity Escalation, 17% slower before, 4% slower after
    i7 3770 and R9 290x: Ave Framerate: 56.2. Ave CPU Framerate: 59.7 FPS.
    i3 10105F and R9 290x: Ave Framerate: 47.0. Ave CPU Framerate: 47.1 FPS
    PL+ i3 10105F and R9 290x: Ave Framerate: 54.0. Ave CPU Framerate: 54.9 FPS

    *Assetto Corsa 6% faster before, 15% faster after
    i7 3770 and R9 290x: 145 Ave FPS, 69 Min, 189 Max, Variance=4 CPU=72%
    i3 10105F and R9 290x: Ave 154 FPS, 50 Min, 203 Max, Variance=15 CPU=89%
    PL+ i3 10105F and R9 290x: Ave 171 FPS, 33 Min, 233 Max, Variance=16 CPU=90%

    *Assassin's Creedy Odyssey 12% slower, 6.5% slower after
    i7 3770 and R9 290x: FPS: 46, Min 21, Max 75
    i3 10105F and R9 290x: FPS: 38, Min 22, Max 54
    PL+ i3 10105F and R9 290x: FPS: 43, Min 23, Max 67
     
    Elmf likes this.
  6. Butcher9_9

    Butcher9_9 Member

    Joined:
    Aug 5, 2006
    Messages:
    2,496
    Location:
    Perth , St James
    Can I ask why the side grade? 3770 machine dead or something?

    4 core CPUs are pretty much on life support for gaming these days.
     
    Last edited: May 5, 2021
    Dr Evil likes this.
  7. OP
    OP
    munted123

    munted123 Member

    Joined:
    Aug 21, 2001
    Messages:
    61
    Ok more benchmarks because obviously I was bored.
    So we've got two different GPUs I've tested to see if the R9 290X is a bottleneck for a i7 3770 / i3 10105F, the other GPU is a 5700XT.

    While I was testing the 5700XT in the i7 3770 and i3 10105F I chucked the R9 290X in my main PC, a Ryzen 3600X. However because my RAM is getting RMA'd at the moment I'm running single channel 16GB 3200MHz, this seems to be quite a bit slower than dual channel than 32GB 3200MHz. For example before my RAM died I did an Ashes of the Singularity Escalation benchmark, result was Ave FPS 98.6 with dual channel, 73.4 FPS with single channel! With Assassin's Creedy Odyssey, result was Ave FPS 77 with dual channel and 60 FPS with single channel. So about 25% slower for both benchmarks...

    I also had a couple of results saved from my Ryzen 2600 with the 5700XT so I added them in because why not?

    I mostly bought it for Mechwarrior Online which is a heavily CPU limited game but doesn't make much use of more than 8 threads. You'll see the results at the end of this post.

    To be honest not many games I play use more than 8 threads and mostly really only care about 1 or 2 fast threads, the other threads are most going to be well less than 100% utilization.
    Last year a Techpowerup review compared the 4 core, 8 thread Ryzen 3300x to the 6 core, 12 thread 3600, on average across 10 games the 3300x was only 1% slower.
    https://www.techpowerup.com/review/amd-ryzen-3-3300x/15.html

    So yeah that was my thinking. I have to admit I was expecting a bigger increase going from a 3770 to a 10105F though. When I upgraded from a Ryzen 2600 to a 3600x which is only 1 generation it was a big uptick in gaming performance. You can see in the results below in Ashes of the Singularity Escalation the 3600x was 20% than the 2600 and in Assassin's Creedy Odyssey it was 25% faster. A mate recently went from a Ryzen 1600 to a 3600 and the difference was huge for 2 Ryzen generations, in Mechwarrior Online he was getting more than double the frame rate. So I would have thought that 7 generations of Intel Core CPUs would be a really big performance difference but apparently not, 7 generations of Intel improvements is less than 1 generation of AMD improvement.

    Results with R9 290X GPU at 1080p

    Uniengine Superposition less than 10105F 2% faster than 3770
    i7 3770 and R9 290x: Score: 8174, FPS Min: 51.47, Ave: 61.14, Max: 76.57. GPU Min 29 deg, Max 69 deg. GPU Utilization: 100%
    PL+ i3 10105F and R9 290x: Score: 8339, FPS Min: 52.28, Ave: 62.37, Max: 77.06. GPU Min 35.0 deg, Max 74.0 deg. GPU Utilization: 100%
    3600X and R9 290x: Score: 8322, FPS Min: 52.23, Ave: 62.25, Max: 77.12. GPU Min 49.0 deg, Max 72.0 deg. GPU Utilization: 100%

    3dMark Time Spy, 10105F 3% faster after than 3770
    i7 3770 and R9 290x: Score 3958. Graphics score 3998. CPU score 3748.
    PL+ i3 10105F and R9 290x: Score 4091. Graphics score 4049. CPU score 4353.
    3600X and R9 290x: Score 4270 Graphics score 4004. CPU score 6861

    Company of Heroes 2, 10105F 2% faster after than 3770
    i7 3770 and R9 290x: Min 42.57, Max 103.28, Ave 65.23
    PL+ i3 10105F and R9 290x: Min 43.61, Max 108.46, Ave 69.06
    3600X and R9 290x: Min 51.79, Max 122.65, Ave 78.36

    Ashes of the Singularity Escalation GPU focused 10105F 4% slower after than 3770
    i7 3770 and R9 290x: Ave Framerate: 56.2. Ave CPU Framerate: 59.7 FPS.
    PL+ i3 10105F and R9 290x: Ave Framerate: 54.0. Ave CPU Framerate: 54.9 FPS
    3600X and R9 290x: Ave Framerate: 63.0. Ave CPU Framerate: 76.3 FPS

    *Assetto Corsa 10105F 15% faster than 3770
    i7 3770 and R9 290x: 145 Ave FPS, 69 Min, 189 Max, Variance=4 CPU=72%
    PL+ i3 10105F and R9 290x: Ave 171 FPS, 33 Min, 233 Max, Variance=16 CPU=90%
    3600X and R9 290x: Ave 173 FPS, 38 Min, 222 Max, Variance=4 CPU=61%

    *Assassin's Creedy Odyssey 10105F 6.5% slower than 3770
    i7 3770 and R9 290x: Ave FPS: 46, Min 21, Max 75
    PL+ i3 10105F and R9 290x: Ave FPS: 43, Min 23, Max 67
    3600X and R9 290x: Ave FPS: 50, Min 27, Max 86

    *Mechwarrior Online Start walk (average of 3 runs) 10105F 18% faster than 3770
    i7 3770 and R9 290x: Ave FPS: 51.67
    PL+ i3 10105F and R9 290x: Ave FPS: 71.67
    3600X and R9 290x: Ave FPS: 75.46

    *Mechwarrior Online Gauntlet walk (average of 3 runs) 10105F 25% faster than 3770
    i7 3770 and R9 290x: Ave FPS: 84.33
    PL+ i3 10105F and R9 290x: Ave FPS: 111.41
    3600X and R9 290x: Ave FPS: 120.434


    Results with 5700XT GPU at 1080p

    *Uniengine Superposition 10105F 11% faster than 3770
    3770 5700XT: Score: 14448, FPS Min: 72.77, Ave: 108.06, Max: 146.68. GPU Min N/A deg, Max N/A deg. GPU Utilization: N/A
    PL+ i3 10105F and 5700XT: Score: 16239, FPS Min: 82.94, Ave: 121.46, Max: 154.04. GPU Min N/A deg, Max N/A deg. GPU Utilization: N/A
    3600x 5700XT: Score: 16588, FPS Min: 91.12, Ave: 124.07, Max: 153.96. GPU Min N/A deg, Max N/A deg. GPU Utilization: N/A

    *3dMark Time Spy 10105F 4% faster than 3770
    3770 5700XT: Score 7544. Graphics score 9205. CPU score 3730
    PL+ i3 10105F and 5700XT: Score 7840. Graphics score 9090. CPU score 4408
    2600 5700XT: 8216. Graphics score 8896. CPU score 5733
    3600x 5700XT: Score 8708. Graphics score 9164. CPU score 6794

    *Company of Heroes 2 10105F 20% faster than 3770
    3770 5700XT: Min 36.82, Max 60.22, Ave 48.56
    PL+ i3 10105F and 5700XT: Min 46.03, Max 103.17, Ave 66.99
    3600x 5700XT: Min 50.45, Max 120.76, Ave 77.54

    *Ashes of the Singularity Escalation GPU focused 10105F 2% slower than 3770
    3770 5700XT: Ave Framerate: 56.0. Ave CPU Framerate: 56.0 FPS
    PL+ i3 10105F and 5700XT: Ave Framerate: 54.9. Ave CPU Framerate: 53.9 FPS
    2600 5700XT Ave Framerate: 79.5. Ave CPU Framerate: 79.9 FPS.
    Dual Channel 3600x 5700XT: Ave Framerate: 98.6. Ave CPU Framerate: 98.8 FPS
    Single Channel 3600x 5700XT: Ave Framerate: 73.4. Ave CPU Framerate: 73.4 FPS

    *Ashes of the Singularity Escalation CPU focused forgot to run on 10105F :(
    3770 5700XT: Ave frametime 19.2 FPS, Ave CPU frametime, 19.2 FPS.
    3600x 5700XT: Ave frametime 34.6 FPS, Ave CPU frametime, 34.6 FPS.

    *Assetto Corsa 10105F 10% faster than 3770
    3770 5700XT: Ave 151 FPS, 12 Min, 191 Max, Variance=7 CPU=89%
    PL+ i3 10105F and 5700XT: Ave 167 FPS, 29 Min, 213 Max, Variance=13 CPU=90%
    3600x 5700XT: Ave 227 FPS, 32 Min, 294 Max, Variance=9 CPU=85%

    *Assassin's Creedy Odyssey 10105F 2% faster than 3770
    3770 5700XT: FPS 44, Min 16, Max 70
    PL+ i3 10105F and 5700XT: FPS: 45, Min 18, Max 69
    2600 5700XT: FPS 57, Min 18, Max 90
    Dual Channel 3600x 5700XT: FPS 77, Min 18, Max 136
    Single Channel 3600x 5700XT: FPS 60, Min 24, Max 89

    *Mechwarrior Online Start walk (average of 3 runs) 10105F 25% faster than 3770
    3770 and 5700XT: Ave FPS: 50.50
    PL+ i3 10105F and 5700XT: Ave FPS: 67.12
    3600X and 5700XT: Ave FPS: 74.93

    *Mechwarrior Online Gauntlet walk (average of 3 runs) 10105F 18% faster than 3770
    3770 and 5700XT: Ave FPS: 86.50
    PL+ i3 10105F and 5700XT: Ave FPS: 104.90
    3600X and 5700XT: Ave FPS: 119.68


    Conclusion
    Well considering I bought this CPU mostly to play Mechwarrior Online it's a pretty significant improvement of around 20% better FPS compared to my old 3770 so that's great. It is not so good that the game is so poorly written that upgrading from a R9 290X to a 5700XT GPU which is twice as fast...makes no performance difference in this game, amazing.

    Also after all that testing I learned that the 3700 vs 10105F CPU with a R9 290X, the 10105F is 2% faster in the gaming benchmarks I ran. The 3700 vs 10105F CPU with a 5700XT, the 10105F is 12% faster.
    And finally the AMD 3600X is 9% faster than the Intel 10105F with a R9 290X and 16% faster with a 5700XT although the 3600X is badly gimped because I only have one memory stick installed in my main PC at the moment. If Intel wasn't so stingy and didn't lock the memory to 2666Mhz with a B460 board it'd be even closer. I suppose it'd be interesting to test a 4C/8T 3300X but I don't have one.
     
  8. Annihilator69

    Annihilator69 Member

    Joined:
    Feb 17, 2003
    Messages:
    6,087
    Location:
    Perth
    Can you run some CPU tests to compare also?
    Cinebench or even Intel Burn Test/SuperPi.

    It's a bit of a side grade really.
    Even though memory would be from 1600mhz to 2666mhz, I'm sure the latency of the 1600mhz would be way lower, so the overall speed of the ram would be quite similar. If you had 3600mhz ram it would be a bit difference.

    CPU Frequency wise, 10th gen is a bit better clock for clock over 3rd gen, but we're talking say 20% here, and this is with the high end chips, so the lower end ones with gimped cache/memory, it's probably 10% tops.

    It'd be interesting to see what frequency the 10th gen actually runs at.
    Turbo boost these days, says up to 4.4Ghz, but this could be for say 30 seconds before it goes down.

    This is where you've found the power unlocking options in BIOS, this would hold turbo for higher and longer, assuming cooling is sufficient as it's essentially auto overclocking.

    Manually overclocking would be best, but I'm not sure if your system would support it.
    Like for me I have a i9-10850k with max turbo of 5.2Ghz, but it only does that when it's under low core load, high load and it only goes 4.8Ghz.

    Manual over clocking of the profiles, means that I can let the CPU go 5.2Ghz when all cores are under load which then adds up to performance.

    Even if games only use say 4 CPU's, having 6-8 CPU is useful.
    Think about it these days, 1 Core just for OS, 1 Core for Browser/AV/Background tasks, 4 remainder cores for the actual game which get all used.

    Also check your power profile settings in Windows.
    It can make a difference between power saver vs max performance, as it will affect the power budget/frequency.
     
    Last edited: May 5, 2021
  9. Butcher9_9

    Butcher9_9 Member

    Joined:
    Aug 5, 2006
    Messages:
    2,496
    Location:
    Perth , St James
    The thing with Intel CPUs is that the only big change in many many generations was the clock speed. Very little in the way of IPC gains.

    With most CPUs that's not too much of an issue as you would get a big clock bump but in your case you went from a i7 to an i3 and in that case you did not gain cores or clock so performance was about the same.

    Ryzen had a big IPC uplift from Zen 1 to Zen 2 and then to Zen 3 as well as a small clock speed increase too so a win all round.
     
  10. Adona

    Adona Member

    Joined:
    Apr 14, 2018
    Messages:
    75
    Location:
    Strathpine, Brisbane
    I must say I'm not that happy with my 10700K
     
  11. demiurge3141

    demiurge3141 Member

    Joined:
    Aug 19, 2005
    Messages:
    2,255
    Location:
    Melbourne 3073
    4 cores holding it back
     
  12. Adona

    Adona Member

    Joined:
    Apr 14, 2018
    Messages:
    75
    Location:
    Strathpine, Brisbane
    cores or the engineering mob at Intel? kinda regret the purchase and feel a ryzen would have been better choice
     
  13. Myne_h

    Myne_h Member

    Joined:
    Feb 27, 2002
    Messages:
    11,421
    Ok, I'm still seeing nothing but gaming benchmarks - unless I've missed something.

    Perhaps I was unclear.

    We can't figure out why your CPU is slow if we can't compare it to just CPUs.

    Run a CPU benchmark of some kind.

    And post the cpuz readout of the ram tab.

    My guess at this point is that it's running single channel and/or ultra shit timings.

    But it's a guess. If it is that, it will be shit in cpu/ram benchmarks.
     
    Last edited: May 5, 2021
  14. de_overfiend

    de_overfiend Member

    Joined:
    Jul 12, 2001
    Messages:
    2,495
    Location:
    Gold Coast
    You cant really compare a 10 year old i7 to a new i3.

    No surprise that the newer i3 matches the performance of an older gen i7. The 3770 was the cream of the crop at the time and to compare it to a latest gen i3 just shows how much the budget cpus have improved since then.

    even with the chipset advances your benchmarks are no surprise to me

    see here for a basic comparison:
    https://cpu.userbenchmark.com/Compare/Intel-Core-i7-3770-vs-Intel-Core-i3-10105F/1979vsm1482345

    https://www.cpu-monkey.com/en/compare_cpu-intel_core_i3_10100f-1762-vs-intel_core_i7_3770-322
     
    Last edited: May 5, 2021
  15. Ripley

    Ripley Member

    Joined:
    Aug 12, 2001
    Messages:
    582
    Location:
    The dark side of the moon
    <begin anecdotal evidence>

    I'm quite happy with my step up from an i7 3770k to a an i7 10700k. That might have something to do with the big step up in memory, and maybe even the move from SATA SSD's to 3400MB/s NVMe SSD's. It might also be because I only do overclocking via Intel speed step, so I'm not seeking out every last possible clock cycle. I figure as big an improvement will come from a step up from the FE1080, when prices finally become *not stupid*, but that's a few years away at least.

    All in all, eight years from a CPU ain't bad at all, especially given that I ask five, but I can see how some might be disappointed that a quantum leap didn't happen. This system is definitely smoother. Games that previously had CPU or IO stutter are smooth as silk. Not that I'm not trying to run Cyberpunk 20-whatever-it-was at silly resolutions. I'm also not running benchmarks trying to justify my purchase. I have to return this motherboard for RMA due to a non-functioning USB port soon, so I'm wondering how much I'll notice the return slowdown. The old system sits there waiting. I'll be finding out soon enough.

    <end anecdotal evidence>
     
    Last edited: May 6, 2021
  16. groovetek

    groovetek Member

    Joined:
    Oct 19, 2003
    Messages:
    3,243
    Location:
    Melbourne
    I too went from a 3770k @ 4.8ghz all-core, to a 10700k @ 5.2ghz all-core.

    Imo for gaming, it has been an improvement, but not something that really justifies 8 years difference.
     
  17. de_overfiend

    de_overfiend Member

    Joined:
    Jul 12, 2001
    Messages:
    2,495
    Location:
    Gold Coast
    OP went from 10+ year old i7 to a brand new i3... just to clarify

    your posts are going from old i7 to new i7

    OP is stating his new i3 isnt much faster than his old i7...
     
  18. groovetek

    groovetek Member

    Joined:
    Oct 19, 2003
    Messages:
    3,243
    Location:
    Melbourne
    I was adding to Ripley's comment effectively, and agreeing with the OP to some extent that the jump isn't as significant, not for gaming anyway, as one may have originally expected.

    Upgrading from a 3770 to 10100 would therefore be much less significant.

    His numbers don't quite like up on the 10100 with what I'd expect however so maybe worth looking into that before drawing conclusions.

    The CPU Test in 3DMark TimeSpy ain't a bad way of quickly gauging comparative CPU performance in gaming.
     
  19. Sunder

    Sunder Member

    Joined:
    Apr 26, 2012
    Messages:
    4,283
    My bet is that power management and thermal management is gimping performance. Looking at benchmarks, on pure CPU performance, you should be looking at 20-30% gains depending on task.

    My work issued me a 8665u based laptop. Not overclockable CPU. Dell's BIOS is conservative as. It was thermally throttling during office work, and bouncing off the TJmax with VMs running. As in dropping all cores back to 1.9Ghz for a few seconds every few seconds.

    A lick of liquid metal and dropping 120mv in Throttle stop and I can't get past 60*C now, and it will run at 4+Ghz all day. (Still power limited, extended the short and long burst limits in XTU, but I think either Intel or Dell have set some hard limits)
     

Share This Page

Advertisement: