10 Core/5700 XT versus 8 Core/5500XT Benchmark Results

[ThornTO]

Hey all, got my 10 Core/5700XT and here are the numbers versus 8 Core/5500XT, which I lifted from MaxTechs video below.

MaxTech: 8 Core @ 3.8, 5500XT 8GB, 32GB Ram
Me: 10 Core @ 3.6, 5700XT 16GB, 40 GB Ram

Geekbench 5.23 Single / Multi:
8 Core: 1273 / 8952
10 Core: 1264 (-1%)/ 7454 (-17%)

That surprised me. I am not sure exactly how Max Tech tested. Numbers above for me are average of 3 consecutive runs with auto (default) fan speed. I also ran again with fans preset to max and starting from a package temp of 40 Degc. Here are all of those runs:
1252 / 7501
1277 / 7339
1292 / 7523
avg = 1264 / 7454
max fan, start at 40 degc pkg: 1346 / 7544

Unless I am missing something, no straight advantage here for the 10 core v. 8 core? I am most surprised by the multicore. How can it be worse?

Geekbench 5.23 Metal:
8 Core 5500XT: 41,727
10 Core 5700XT: 55,973 (+34%)

Again, I averaged 3 consecutive runs, then did a max fan.
56,682
56,394
54,842
avg = 55,973
max fan / 40 Degc: 55755

Cinebench R20.06:
8 Core 5500XT: 4,968
10 Core 5700XT: 5,413 (+9%)

Run by run:
5448
5397
5394
max fan = 5588

Hopefully I didn't make any typos.

Graphics seems decent but the CPU, per Geekbench at least, is "worse"? Again please help if I am missing something basic.

Cheers.

 

[wavesm]

Hey all, got my 10 Core/5700XT and here are the numbers versus 8 Core/5500XT, which I lifted from MaxTechs video below.

MaxTech: 8 Core @ 3.8, 5500XT 8GB, 32GB Ram
Me: 10 Core @ 3.6, 5700XT 16GB, 40 GB Ram

Geekbench 5.23 Single / Multi:
8 Core: 1273 / 8952
10 Core: 1264 (-1%)/ 7454 (-17%)

That surprised me. I am not sure exactly how Max Tech tested. Numbers above for me are average of 3 consecutive runs with auto (default) fan speed. I also ran again with fans preset to max and starting from a package temp of 40 Degc. Here are all of those runs:
1252 / 7501
1277 / 7339
1292 / 7523
avg = 1264 / 7454
max fan, start at 40 degc pkg: 1346 / 7544

Unless I am missing something, no straight advantage here for the 10 core v. 8 core? I am most surprised by the multicore. How can it be worse?

Geekbench 5.23 Metal:
8 Core 5500XT: 41,727
10 Core 5700XT: 55,973 (+34%)

Again, I averaged 3 consecutive runs, then did a max fan.
56,682
56,394
54,842
avg = 55,973
max fan / 40 Degc: 55755

Cinebench R20.06:
8 Core 5500XT: 4,968
10 Core 5700XT: 5,413 (+9%)

Run by run:
5448
5397
5394
max fan = 5588

Hopefully I didn't make any typos.

Graphics seems decent but the CPU, per Geekbench at least, is "worse"? Again please help if I am missing something basic.

Cheers.

I think your problem is with the RAM; you are mixing modules I am sure their clock has dropped.

Try only with the 32GBs you added.

If you see the lowest Geekbench most are or only 8GB or the ones that added mem but lower clock.

 

[opeter]

Well, the 8-core CPU has faster core speed, than the 10-core CPU. That is, why the difference.

It is similar, when you look at th results of the 4-core Mac mini (that has a 3.6 GHz CPU) vs. the 6-core CPU (3 GHz). At least in single core speed.

 

[oneafour]

is the Geekbench test making use of all 10 cores? This is upsetting.

 

[ThornTO]

Hmm, The system reports all are good and at 2667. I have two more 16Gb sticks coming tomorrow, So I will rerun with 32GB, as well as 64Gb.

 

[wavesm]

Hmm, The system reports all are good and at 2667. I have two more 16Gb sticks coming tomorrow, So I will rerun with 32GB, as well as 64Gb.
View attachment 943489

Let us know how it goes.

 

[KrazyKanuck]

Hmm, The system reports all are good and at 2667. I have two more 16Gb sticks coming tomorrow, So I will rerun with 32GB, as well as 64Gb.
View attachment 943489

I believe that you've dropped dual channel in that configuration. To keep dual channel you need to have a matching pair in DIMM 0 and a matching pair in DIMM 1. But on these 2020 iMacs that seems to lower the speed if you're also using a pair of Apple stock memory. Try with your pair of 16GBs in the same DIMM. I had the same problem but with my i7, I just couldn't get the correct Crucial RAM to cooperate with Apple's SK Hynix RAM. I ended up adding four sticks of the same Crucial RAM to achieve dual channel and 2667. Your single core seems about right (maybe a little low); your multi-core should be nearer 10,000. For comparison my i7 w/5500X and 32GB got about 1390 SC and 8990 MC.

 

[RyanXM]

Slots 1 and 3 should match.
Slots 2 and 4 should match.

This is causing your lower scores.

I have a 10-core that will be here on Wednesday and I will have a detailed thread with all the different RAM configurations and speeds, etc.

 

[Whiteman007]

Yeah the 10 core I have seen benchmarks of has been around 9800 about a 10% boost. Something must be messed up on yours seems like to be the ram not being duel channel

 

[NorCalZman]

Slots 1 and 3 should match.
Slots 2 and 4 should match.

This is causing your lower scores.

I have a 10-core that will be here on Wednesday and I will have a detailed thread with all the different RAM configurations and speeds, etc.

Thanks for doing all that. Please make it a new thread as I do not want to miss your findings. My iMac will be here next week and I wont be ordering any RAM until I see your results.

 

[ThornTO]

Yup, that was it. Sorry for causing anyone alarm! I am remembering all this now. I had 2x16 on hand when it arrived yesterday so I just through them in, and checked to see system saw all at full speed. My extra 2x 16 are coming tomorrow, and that is what I will really run. 64 as 4x 16.

I redid everything with just 2x 16. And everything got better. I know these are synthetic benchmarks, but I still question the value of upgrading to the 10 core for 3-4% better. Like to see these numbers for the 8 core but all else same.

MaxTech: 8 Core @ 3.8, 5500XT 8GB, 32GB Ram
Me: 10 Core @ 3.6, 5700XT 16GB, 32 GB ram as 2X 16 (CHANGED from above)

Geekbench 5.23 Single / Multi:
8 Core: 1273 / 8952
10 Core: 1317 (+3%) / 9324 (+4%)

Numbers above for me are average of 3 consecutive runs with auto (default) fan speed. I also ran again with fans preset to max and starting from a package temp of 40 Degc. Here are all of those runs:
1259 / 9340
1355 / 9329
1336 / 9303
avg = 1317 / 9324
max fan, start at 40 degc pkg: 1350 / 9332

Geekbench 5.23 Metal:
8 Core 5500XT: 41,727
10 Core 5700XT: 57,053 (+37%)

Again, I averaged 3 consecutive runs, then did a max fan.
57617
55968
57573
avg = 57,053
max fan / 40 Degc: 55699

Cinebench R20.06:
8 Core 5500XT: 4,968
10 Core 5700XT: 5,447 (+10%)

Run by run:
5520
5426
5394
max fan = 5584

There you go. Id like to see these numbers on an 8core / 5700XT rig. Im kinda thinking the 10 core isn't worth it.

Sorry again for the initial screw up.

Cheers.

 

[quagmire]

The Metal scores for the 5700 XT seem disappointing compared to the Vega 48. Only ~10% faster?

Making me rethink my plan....

 

[GreenDice]

Im kinda thinking the 10 core isn't worth it.

For a total of $900, “10 core + 5700XT 16GB” is hard to justify based on this benchmark. It is probably a better story for other usages but I doubt that it is $900 better than the 8 core + 5500XT combo.

 

[elicash]

For a total of $900, “10 core + 5700XT 16GB” is hard to justify based on this benchmark. It is probably a better story for other usages but I doubt that it is $900 better than the 8 core + 5500XT combo.

Seems like either something is up with the machine during the tests, or it's like you say. Enough to make me one to put a hold on my order for clarity's sake.

 

[KrazyKanuck]

Seems like either something is up with the machine during the tests, or it's like you say. Enough to make me one to put a hold on my order for clarity's sake.

Agreed. I'd like to see a little more in the way of real world comparisons as well, especially between the three options for the tier-3 model. The figures in the above tests (post RAM troubleshooting) seem in line with what I've been seeing. The i9 was never going to offer jaw-dropping improvements over the i7. But if those extra cores can save you time/make you money, then sure. But a very small portion of users would really fall into this category. Buying an i9 for "future proofing" is silly. By the time the i7 becomes noticeably underwhelming in daily life it will be time for a new machine regardless, and much will have changed. I can see a case for the upgraded graphics, however, and hence my desire for some more comparisons. The more I consider it, the more I think the more meaningful comparison is between the 5500 XT and the 5700. The 5700 XT is something of a niche upgrade. For gamers (or those who have been waiting for the day when they could game decently on an iMac (poor Azrael 😉), I get it; for really demanding video work, I get it; for those who want the best, I get it. But for "future proofing" or for most of the rest of us, I don't think so. 20% more $ for an i9 for a 10% boost in multi-core performance isn't worth it; 20% more for a 35% boost in graphics might be, but most of us will never realize that gain before upgrading to a new machine becomes too worthwhile. But if the 5700 can offer a 27-30% boost in performance for a 10% increase in cost, that's pretty sweet.

I should also add, that what makes so much of this fun is the absurd nature of debating these upgrades. Many of us (well, at least me) would happily spend the equivalent of a 5700 upgrade on a trip to the liquor store or a couple of nights out at the pub. When that's how the money flows it seems like a perfectly reasonable expense; but when it's a 5500XT vs. a 5700 vs. a 5700 XT it causes no end of existential angst and agony.

 

[KrazyKanuck]

Yup, that was it. Sorry for causing anyone alarm! I am remembering all this now. I had 2x16 on hand when it arrived yesterday so I just through them in, and checked to see system saw all at full speed. My extra 2x 16 are coming tomorrow, and that is what I will really run. 64 as 4x 16.

I redid everything with just 2x 16. And everything got better. I know these are synthetic benchmarks, but I still question the value of upgrading to the 10 core for 3-4% better. Like to see these numbers for the 8 core but all else same.

MaxTech: 8 Core @ 3.8, 5500XT 8GB, 32GB Ram
Me: 10 Core @ 3.6, 5700XT 16GB, 32 GB ram as 2X 16 (CHANGED from above)

Geekbench 5.23 Single / Multi:
8 Core: 1273 / 8952
10 Core: 1317 (+3%) / 9324 (+4%)

Numbers above for me are average of 3 consecutive runs with auto (default) fan speed. I also ran again with fans preset to max and starting from a package temp of 40 Degc. Here are all of those runs:
1259 / 9340
1355 / 9329
1336 / 9303
avg = 1317 / 9324
max fan, start at 40 degc pkg: 1350 / 9332

Geekbench 5.23 Metal:
8 Core 5500XT: 41,727
10 Core 5700XT: 57,053 (+37%)

Again, I averaged 3 consecutive runs, then did a max fan.
57617
55968
57573
avg = 57,053
max fan / 40 Degc: 55699

Cinebench R20.06:
8 Core 5500XT: 4,968
10 Core 5700XT: 5,447 (+10%)

Run by run:
5520
5426
5394
max fan = 5584

There you go. Id like to see these numbers on an 8core / 5700XT rig. Im kinda thinking the 10 core isn't worth it.

Sorry again for the initial screw up.

Cheers.

Many thanks for running these tests and sharing the results with us!

 

[Jax44]

Agreed. I'd like to see a little more in the way of real world comparisons as well, especially between the three options for the tier-3 model. The figures in the above tests (post RAM troubleshooting) seem in line with what I've been seeing. The i9 was never going to offer jaw-dropping improvements over the i7. But if those extra cores can save you time/make you money, then sure. But a very small portion of users would really fall into this category. Buying an i9 for "future proofing" is silly. By the time the i7 becomes noticeably underwhelming in daily life it will be time for a new machine regardless, and much will have changed. I can see a case for the upgraded graphics, however, and hence my desire for some more comparisons. The more I consider it, the more I think the more meaningful comparison is between the 5500 XT and the 5700. The 5700 XT is something of a niche upgrade. For gamers (or those who have been waiting for the day when they could game decently on an iMac (poor Azrael 😉), I get it; for really demanding video work, I get it; for those who want the best, I get it. But for "future proofing" or for most of the rest of us, I don't think so. 20% more $ for an i9 for a 10% boost in multi-core performance isn't worth it; 20% more for a 35% boost in graphics might be, but most of us will never realize that gain before upgrading to a new machine becomes too worthwhile. But if the 5700 can offer a 27-30% boost in performance for a 10% increase in cost, that's pretty sweet.

I should also add, that what makes so much of this fun is the absurd nature of debating these upgrades. Many of us (well, at least me) would happily spend the equivalent of a 5700 upgrade on a trip to the liquor store or a couple of nights out at the pub. When that's how the money flows it seems like a perfectly reasonable expense; but when it's a 5500XT vs. a 5700 vs. a 5700 XT it causes no end of existential angst and agony.

I am trying to decide between the 5500XT and the 5700. I plan on getting the 8 core i7.

 

[smbu2000]

The 10core multicore GB5 scores tend to be in the 9000-10000 range from a quick search.

https://browser.geekbench.com/v5/cpu/search?utf8=✓&q=imac20,2

 

[Hexley]

The Metal scores for the 5700 XT seem disappointing compared to the Vega 48. Only ~10% faster?

Making me rethink my plan....

Future macOS update may solve that.

 

[quagmire]

Future macOS update may solve that.

Maybe plus the Geekbench scores overall seem off. Initial benchmarks of the 10700 had it in the 7000's multi-core. No way is the 10700 slower than the 9900K with the 10700 being 200 Mhz faster base clock and having the same turbospeed. Cinebench is showing more accurate results than Geekbench.

 

[SkiHound2]

Agreed. I'd like to see a little more in the way of real world comparisons as well, especially between the three options for the tier-3 model. The figures in the above tests (post RAM troubleshooting) seem in line with what I've been seeing. The i9 was never going to offer jaw-dropping improvements over the i7. But if those extra cores can save you time/make you money, then sure. But a very small portion of users would really fall into this category. Buying an i9 for "future proofing" is silly. By the time the i7 becomes noticeably underwhelming in daily life it will be time for a new machine regardless, and much will have changed. I can see a case for the upgraded graphics, however, and hence my desire for some more comparisons. The more I consider it, the more I think the more meaningful comparison is between the 5500 XT and the 5700. The 5700 XT is something of a niche upgrade. For gamers (or those who have been waiting for the day when they could game decently on an iMac (poor Azrael 😉), I get it; for really demanding video work, I get it; for those who want the best, I get it. But for "future proofing" or for most of the rest of us, I don't think so. 20% more $ for an i9 for a 10% boost in multi-core performance isn't worth it; 20% more for a 35% boost in graphics might be, but most of us will never realize that gain before upgrading to a new machine becomes too worthwhile. But if the 5700 can offer a 27-30% boost in performance for a 10% increase in cost, that's pretty sweet.

I should also add, that what makes so much of this fun is the absurd nature of debating these upgrades. Many of us (well, at least me) would happily spend the equivalent of a 5700 upgrade on a trip to the liquor store or a couple of nights out at the pub. When that's how the money flows it seems like a perfectly reasonable expense; but when it's a 5500XT vs. a 5700 vs. a 5700 XT it causes no end of existential angst and agony.

That's my take. The difference between an i7 and i9 will be of no consequence to the vast majority of users. The extra cores of the i9 might benefit a select few; but they maybe should be looking at other options. The video card upgrades probably make more sense. But the lowest end i5 version will do most of what most folks do most of the time without breaking a sweat.