Overview Features Coding ApolloOS Performance Forum Downloads Products Order Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.
Please visit our Apollo-Discord Server for support.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Performance and Benchmark Results!

Stephen Jones - Video Review

Gunnar von Boehn
(Apollo Team Member)
Posts 6207
03 Jun 2020 19:19


Stephen Jones did some new Video.
In this video he compared Vampire performance.
       
Many people spoke to us, reporting his video and asked
Why does Steve gets only so poor performance?
       
Frankly, we not know.
Most likely he runs a very old core, and very old GFX drivers
and has not updated his Vampire since years?
Also it looks like some features like FASTIDE are not even enabled.
       
How about we look at his Video now and review the results he gets, and then we see what he could have measures if he would have updates his system and would have enabled all features like FASTIDE correctly?
       
       
Maybe this is also helpful for other people who have not bothered to enable FASTIDE in the last years  or not updates his Vampire and drivers for a number of years.
       
       
So lets look as his scores first.
       


          VAMPIRE500  WARP    Good_Setup_Vampire500
SYSINFO        104    78      166 !
IDE            2.9    4.8    12.0 !
 
EMUTEST                        112 !
WRITEPIXEL      93    83      100 !
SIEVE          103    70      120 !
DHRYSTONE                      110 !
SORT                            125 !
ELIPSETEST      60    52      69 !
MATRIX                          166 !
IMATH                          325 !
MEMTEST                        180 !
TGTEST          38    55      66 !
FMATH          903    516    1060 !
BEACHBALL      1190  1636 !  1393
LINETEST        27    37      41 !
 


     









       
     
Lets see what result you get with correctly enables features and updates.














 


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
03 Jun 2020 22:00


we are currently uploading correct numbers
Please wait with comments until all are uploaded.
Only when you see all results this will make sense.



Vojin Vidanovic
(Needs Verification)
Posts 1916/ 1
03 Jun 2020 22:33


Seems Mr Jones acknowledges he did use the old core etc.
(not in the video, but in the comments section)

Just to put things to perspective, his Warp video is
EXTERNAL LINK 
and it includes the comparison
00:34:52.    Sysinfo CPU - Warp vs Vampire
00:36:50.    Sysinfo Drive speed - Warp vs Vampire
00:38:13.    AIBB v6.5 video tests - Warp vs Vampire

Stéphane Pitteloud
4 days ago
the Vampire V4 and V500 do between 10 and 12 Mb/s. So, the Vampire is in fact faster than the Warp on the IDE.And you use certainly a old core on your Vampire, the 2.12 core achieve more than 140 mips on the Vampire.

Stephen Jones
Cool, maybe I will try that soon.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
03 Jun 2020 22:41


OK we uploaded some good number with good core, and better settings.
 
In my opinion the WARP is an excellent card.
Its is certainly worth to buy.
And if you have an unused 68060 CPU available then please consider getting the Warp.
 
Just like a Harley Davidson Motobike has some quality on its own
and does not need to outrun a SuperSport to convince.
Just the same way the WARP is worth to buy.
 
And this is totally unrelated to any AIBB results,
Using AIBB results to justify the WARP is totally unreasonable in my opinion and is not needed.
 
If you are interested in some real AIBB results
or in correct IDE speed results then you can find them above.
We included some more test than Steve as he accidentally skipped some pretty important ones.
 
But even if the WARP not beats the SuperSport in every category - its worth to get it.
 
Keep the Amiga Spirit!
 
 

Lets us close this thread now!
Please not post.
 
And if someone now has question how to install latest SAGA drivers or how to enable FAST IDE - then please open a new thread.
 


Otto PS
(Needs Verification)
Posts 26/ 1
04 Jun 2020 04:16


I am trying not to be disrespectful because I understand the great effort that Gunnar has made regarding vampire accelerators.

But Gunnar's comparison is at least biased because Stephen is testing a v600 where the latest core has a maximum x11 clock multiplier (There is no public option to download 2.12 faster core for v600). The numbers Gunnar exposes without specifying it are in a core 2.12 with at least a x13 multiplier. The clock difference between the two cores is at least 18.18%. ***
    I wonder what the comparison numbers would be with a v600 with an x11 multiplier ....



Stefano Briccolani

Posts 586
04 Jun 2020 06:27


Steven have a V500 on an Amiga2000 not an v600. Ide on V500 (even on a 2 years old core) is 3 or 4 times faster than depicted on his video.
Maybe he didn't know that IDE speed on vampire is adjustable (to be compatible with device other than cfs) and fast IDE have to be activated to score? Maybe. But before approaching a comparison video with benchmarks and scores (that people assume as scientific facts) he should have asked himself why a product advertised to have a 10mbs ide onboard (V500) on his Amiga does less than 3..


Sean Sk

Posts 488
04 Jun 2020 06:28


Stephen actually has a Vampire 500 (not a V600) which has onboard IDE and means he can set a faster IDE speed using the SAGA driver. I've heard that he may still be using an older core, so he might actually get better performance using a newer core.

EDIT: Posted at exactly the same time as Stefano so apologies if the same information was repeated.


Gunnar von Boehn
(Apollo Team Member)
Posts 6207
04 Jun 2020 08:12


Otto PS wrote:

I am trying not to be disrespectful 

 
Then you should respect that the thread is closed.
 
Otto please make no wrong claims here about what system Stephen was using.
 
Steve uses a V500 in an A2000.
This is the exactly the same setup which was re-tested.
 
The core which was used in the re-test is public available.
Also the GFX drivers and Benchmark version tools are public available.
 
 
Regarding the Warp, the WARP is a nice product which stands is ground, and which we highly recommend owners of the 68060 to buy.
 
This post was not to start more comparison of WARP, and we will not support such.
 
 
The goal of the post is to help Stephen Jones with reference values.
I have the feeling that some things went wrong in his video.
Whether this happened on purpose or by accident I not want to guess.
But in all cases explaining how to correctly setup such comparison could not be wrong and maybe even helps future videos?
 
 
Stephen makes Amiga videos in which he compares Amiga accelerator products and maybe he even  generates some money revenue over this.
This is great and I find his videos look nice and are well spoken.
 
What I have problem to understand is why Stephen did some things in his setup like he did?
 
Lets give some examples:
A) FASTIDE
We all know that FASTIDE is a major feature of the Vampire
and this feature was always well advertised on the Vampire.
As you will know all modern PCs support faster IDE modes,
also for the Amiga 4000 this was planned by Commodore.
Also the Vampire follow Commodore plans and supports several faster IDE settings.
You can enable FASTIDE with 1 click in the config.
If FASTIDE is enabled you should get around 10-13 MB/sec speed.
With some HDD or CF models you can even get up to 18-19MB/sec.
 
Stephen did put a lot emphasis on the IDE performance of the accelerators in his video.
This makes it for me hard to understand why he not not enabled FastIDE on his Vampire?
 
 
There are a few more shortcoming in his video.
For sure that making such nice video was a hell of a job and Stephen for did spend many hours doing this.
As we all know the Vampire core and the GFX drivers are regularly updated.
What I have problems to understand and find very unfortunate
is that Stephen seems to run a core maybe 3 years old and he did not spend a few minutes time to update it before starting this very extensive benchmark video.
 
 
How the benchmarks were done.
If you understand how AIBB works then you know that AIBB comes with a database of results for A600, A1200 and A4000.
All the bars that AIBB renders are based on this database.
What Stephen did "probably accidentally"? select a resolution not stored in AIBB database. This means AIBB did render bars - but all of them are wrong - as AIBB has no reference stored.
 
I see the purpose of his videos to help people to understand stuff better.  This is a great goal.
I think if you do this then you also have the responsibility to setup the system correctly and to show people real results.
 
Here Stephen selected an unsupported resolution
and because of this all rendered bars the AIBB GFX test are all totally wrong. I'm not sure that showing wrong numbers will help people.
 
There is a number of more points a good benchmark person would have considered and taken care of.
For example with wrong setup of system patches "e.g. Fastblit" you can screw up tests like LineDraw.
Also AIBB allows you to select several FPU modes for its test.
Here the careful tester has the option to influence the test to the better or worse results. E.g by selecting 2 different FPU routines when comparing 2 products you can accidentally or on purpose "influence" the results.

 
 
I spoke a lot about technical details here.
The moral of the story is.
Colored Benchmarks Bars are always nice to look at.
But the viewer always needs to take care on how the tests were produced.
 
Before this thread was posted here, some people tried to point the benchmark mistakes to Stephen directly and he seemed to be not interested. I'm not sure about the intension of Stephen video.
Before this video Stephen asked me for cooperation.
And I told him like all other resellers that we sell our V4 currently by ourselves. And now he is making this video which is full of mistakes, this leaves me a little bit puzzled.
 
But all is good, I just wrote this above not in disrespect to Stephen but to make people understand why the numbers are like this.

posts 8