MP All Models Geekbench 5 Extremely Low Score for Mac Pro 5,1

MapleGreen

macrumors member
Original poster
Jun 25, 2019
76
5
Hi there
I've recently installed Geekbench 5
and in the first run I surprised with the extremely low scores and metal benchmark that failed!
I am on High Sierra 10.13.6 and RX580 GPU
It would be great if we compare scores with someone who installed macOS Catalina beta on cMP


this is Geekbench 5 release note :
  • Geekbench 5 is 64-bit only, dropping support for 32-bit processors and operating systems. Geekbench 5 does not include any of the compromises required to run on 32-bit systems. This enables Geekbench 5 to include more ambitious benchmark tests with larger data sets and longer running times.
Screen Shot 2019-09-04 at 10.57.58 PM.png
Screen Shot 2019-09-04 at 10.51.34 PM.png


Screen Shot 2019-09-04 at 10.52.21 PM.png
Screen Shot 2019-09-04 at 10.52.28 PM.png
Screen Shot 2019-09-04 at 10.59.50 PM.png
Screen Shot 2019-09-04 at 10.59.56 PM.png
 
Last edited:
  • Like
Reactions: handheldgames

bsbeamer

macrumors 68030
Sep 19, 2012
2,859
1,393
All MP4,1/MP5,1 CPU's are basically EOL'd by Intel and no longer being updated for security vulnerabilities/mitigation. All mitigation methods provided are from Apple and software level, or require disabling HT to (at least partially) mitigate.

OP - was HT disable performed for this benchmark?

Also will say, this is a bad benchmark and has been for some time. Some people on this forum swear by it, but real world results are often much better (or much different) than what this benchmark will say.
 

MapleGreen

macrumors member
Original poster
Jun 25, 2019
76
5
All MP4,1/MP5,1 CPU's are basically EOL'd by Intel and no longer being updated for security vulnerabilities/mitigation. All mitigation methods provided are from Apple and software level, or require disabling HT to (at least partially) mitigate.

OP - was HT disable performed for this benchmark?

Also will say, this is a bad benchmark and has been for some time. Some people on this forum swear by it, but real world results are often much better (or much different) than what this benchmark will say.
Hyper threading was enabled for this benchmark.
I have no problem in real life performance and I had really good Scores in Geekbench 3
3000 Single-score and 28000 Multi-score Approx.
 

Pressure

macrumors 68040
May 30, 2006
3,945
252
Denmark
Not sure why it should be a surprise. These Mac Pros are from 2009-2012. They are old and probably do not support many of the instructions used to speed up the sub level benchmarks. Just look at the Mac mini (Late 2018) candidly beating our dinosaurs.
 
  • Like
Reactions: Nugget and h9826790

s.m.t.

macrumors 6502
Mar 7, 2010
253
14
Not sure why it should be a surprise. These Mac Pros are from 2009-2012. They are old and probably do not support many of the instructions used to speed up the sub level benchmarks. Just look at the Mac mini (Late 2018) candidly beating our dinosaurs.
That’s what I assumed when I saw the score. They probably updated the tests to best stress more modern hardware. I don’t think the scores between 4 and 5 are comparable.
 

MapleGreen

macrumors member
Original poster
Jun 25, 2019
76
5
Not sure why it should be a surprise. These Mac Pros are from 2009-2012. They are old and probably do not support many of the instructions used to speed up the sub level benchmarks. Just look at the Mac mini (Late 2018) candidly beating our dinosaurs.
I think performing this Benchmark on Catalina shows better Score
hope some one do this and share with us
[doublepost=1567631642][/doublepost]
That’s what I assumed when I saw the score. They probably updated the tests to best stress more modern hardware. I don’t think the scores between 4 and 5 are comparable.
I checked Geekbench site on version 5 section
other results are very low even newer processors
[doublepost=1567631900][/doublepost]
I think performing this Benchmark on Catalina shows better Score
hope some one do this and share with us
[doublepost=1567631642][/doublepost]

I checked Geekbench site on version 5 section
other results are very low even newer processors

this is top scores in version 5 so far

Screen Shot 2019-09-05 at 1.46.49 AM.png
 

tpivette89

macrumors 6502
Jan 1, 2018
416
212
Newark, DE
They are calculating the scores completely differently from Geekbench 4, so the scores between the two benchmarks cannot be compared.

Here is what my dual X5677s (w 48GB RAM) scored yesterday:

Screen Shot 2019-09-04 at 5.46.12 PM.png

Meanwhile, this is the same setup in Geekbench 4:

Screen Shot 2019-09-04 at 5.48.18 PM.png
 

yurc

macrumors regular
Aug 12, 2016
159
77
inside your DSDT
No surprise there. Each time new iteration geekbench comes out, scores are always sliding down on newer version due different baseline CPU measurement.

At circa GB3, dual X5690 scores at 28.000~31.000
At circa GB4, dual X5690 scores at 17.000~20.000

GB5 seems more even worse.

I was start ignoring benchmark result when GB4 comes, because they tend to make me thinking my system are too slow and craving for newer latest and greatest hardware.
 
  • Like
Reactions: MapleGreen

Krevnik

macrumors 68040
Sep 8, 2003
3,318
676
Geekbench scores are relative to a baseline CPU. Primate Labs outlines the baseline on their website:

"Geekbench 4 CPU and Compute scores are calibrated using a Microsoft Surface Book with an Intel Core i7-6600U processor as a baseline with a score of 4,000 points."

"Geekbench 5 CPU scores are calibrated using an Intel Core i3-8100 processor as a baseline."

With Geekbench 5, the baseline is also set to 1000 points, not 4000 like Geekbench 4.
 

MIKX

macrumors 65816
Dec 16, 2004
1,319
487
Japan
Initially quite confusing to be honest .

These grabs are Compute - OpenCL -MSI R 580 8gb.

Geekbench 4
Geekbench 4 COMPUTE benchmark.png


Geekbench 5
Geekbench 5 COMPUTE benchmark.png
 

MIKX

macrumors 65816
Dec 16, 2004
1,319
487
Japan
Definitely odd results with the GB Vs 5 COMPUTE benchmark.
How can there be such a discrepancy between my MSI RX 580 8gb and an RX 480 ) despite the two two GPUs being so related to each other ?
Geekbench 5 COMPUTE COMPARISON benchmark.jpg



Also in the GB Cs, 5 singe/multi benchmark it shows my 32 gb of ECC ram's frequency as being 533 Mhz.
Geekbench 5 COMPUTE RAM frequency mis-read benchmark.jpg
 

smbu2000

macrumors 6502
Oct 19, 2014
268
120
It looks like the compute comparison is using scores from Geekbench 4 and not the upated scores. That RX480 score is about 125k, which is close to the 135k score of your RX580 in Geekbench 4.

It looks to be the same with the CPU comparison scores as well. That i7-4790K multi core score would be close to double that of the 12core E5-2697 V2 in my 6,1 nMP in Geekbench 5.
 

tpivette89

macrumors 6502
Jan 1, 2018
416
212
Newark, DE
Definitely odd results with the GB Vs 5 COMPUTE benchmark.
How can there be such a discrepancy between my MSI RX 580 8gb and an RX 480 ) despite the two two GPUs being so related to each other ?
View attachment 856168


Also in the GB Cs, 5 singe/multi benchmark it shows my 32 gb of ECC ram's frequency as being 533 Mhz.
View attachment 856169
DDR = Double Data Rate, so 533mhz x 2 = 1066. This is normal.
 

yurc

macrumors regular
Aug 12, 2016
159
77
inside your DSDT
Try running GB4 which should make your computer faster.
Yep, also running GB3 is make our dinosaurs cMP more faster lol.

But seriously everyone , don’t take seriously GB score too much. I was remembered some barefeats article in past which minds cMP low score on GB4 results during early GB4 launch and then those article now no longer available, I suspect barefeats retracted them.

Also, GB is performs in cherry picking manner. In order gain higher result, disable any background tasks, any monitoring tools, close all active apps and windows, clear all active program runs on menubar, use triple channel config (if cMP), geekbench should give better 10-20% better results compared compared computer in “normal state” with any background task active.
 

RedTomato

macrumors 601
Mar 4, 2005
4,018
314
.. London ..
Sorry to resurrect this thread - Does anyone know where the Geekbench 4 Mac Benchmarks are? When GB5 came out, it seems they just wiped their GB4 Mac Benchmark charts. The only Mac Benchmark chart I can find now is only for GB5.

For me, I have a list of my macs with GB4 scores, and now I'm considering buying another second hand mac, it would be useful to know its prospective GB4 score to compare with what I already have. I'm sure many people will have their own reasons for wanting to look up the GB4 Mac Benchmarks.
 

itdk92

macrumors 6502
Nov 14, 2016
489
179
Copenhagen, Denmark
Not sure why it should be a surprise. These Mac Pros are from 2009-2012. They are old and probably do not support many of the instructions used to speed up the sub level benchmarks. Just look at the Mac mini (Late 2018) candidly beating our dinosaurs.
Turns out, only on the benchmarks, not on actual professional work luckily..