Kyro 2 - Part 2

Guest · Apr 28, 2001

Those boards are 4 layer PCB's. The final boards will be 6 layered. I can tell you for sure that IMGTEC do not test and hand pick special very overclockable boards for reviews. They don't do things that way. You might not beleive that because it sounds to honest but I have received boards from them in the past for reviews (Kyro 1) and the Kyro 1 I got from them was terrible at overclocking. Loads of people I know who bought the board can clock at least 10mhz too 125mhz and I heard of someone clocking to 135mhz with a blue orb on the board. But mine will clock no more then 2-3mhz without having problems. I know nobody else with such a unoverclockable Kyro 1 card. So thats proof to me that they just send random cards for reviews. They certainly don't have 1 stash of good cards for reviewers and another for actual customers.

Look at this thread at Beyond3d: http://www.beyond3d.com/forum/noncgi/Forum1/HTML/002641.html

Thats Dave Baumann who did the Beyond3d Kyro II preview. He tests the Kyro II Videologic Vivid!XS 32mb for overclocking and gets better results then any review trying to overclock a Herc 4500. The Herc 4500 boards are built on the Videologic reference board (not a GTS board like Nvidia said in there crappy PDF) but obviously the revision of reference design was a little early for the Herc review boards compared to the Vivid!XS boards. When all cards are finalised on a 6 layer PCB we might get decent results. Certainly to say that there's a real possibility these boards are clocked to the max and might brake at any moment is unfiar IMO. IMGTEC don't have a track record for low quality hardware and neither to ST. Some of there boards might have had bugs when they were still refining tile based rendering but the general quality of there products are good (Dreamcast, Naomi 1 and 2, Kyro 1, Sonic Fury etc). No company in there right minds building for the future is going to release a card that brakes allot. The Kyro 1 only overclocked 10mhz for most people and I don't know anyone thats had a board thats broken.

On a final point just to back up what I'm saying I'll take an extract of the paragraph you based your comments on:

"The core does not get hot, but even with improved cooling (a giantic fan blowing air on both the chip and memory chips) we could not get it past 185 MHz"

If the chip was already overclocked and thats why it won't go much higher then the core would be very hot but it isn't, this shows that the reason is not because the board is clocked to the max.

<<<<<Do you know if a mobile Kyro chip solution is in the makings? Seems like an ideal chip for a mobile unit due to its low overhead.>>>>>

PowerVR anounced a Mobile chip a month back (http://www.powervr.com/PressReleases/2001/IMGAndARMAnnounceNewProcessorForMobileAndCommunications.htm). The chip is called PowerVRMBX, some of the interesting features of the PowerVRMBX that Kyro II doesn't have are a HW T&L unit and FSAA4FREE.

Guest · Apr 28, 2001

I've said it before, in Giants with DX8 I loose 10fps when using features that render into textures (shadows and advanced water effects, which are on by defualt) and with DX7 I loose nothing for the same features, so the performance will be much higher when DX8 is fixed.

Surely you see a patern in the benches your talked about, the Kyro II doing great in OpenGL and allot worse in D3D, D3D is actually faster on the Kyro then OpenGL when your not using DX8, again the DX8 problem.

Also I hope you release that Giants has no benchmark. Any benches of Giants is merely using a live framerate counter and taking the lowest framerate number at a specific time in the game and the results there are not an average FPS over a certain time in the game like Q3 or SS.

BTW what review are you talking about?

Edited by Teasy on 04/27/01 10:15 PM.

noko · Apr 28, 2001

So the boards being tested are not the ones that will be sold to the public? This is getting somewhat confusing. I thought the Hercules boards where pretty much the final boards that are going to be shipped.

Heat does not always mean a limitation of a chip design. Other factors such as interference from the circuit design due to inductance and capacitance between the traces can limit the speed of a chip. Looks like heat is not the issue for the Kyro2 but the actual chip is unable to go any faster due to inherit limitations. In a chip which does not have capacitance or inductance limitations heat becomes a issue because as the chip heats up (maybe due to overclocking) causes the resistances to change in the conductive traces, causing voltages to change eventually causing a failure(Usually as resistance goes up the voltage drops, thats why when overclocking and by increasing the voltage helps in the stability of a chip). The Kyro2 in its present form may never go beyond a few mhz in speed.

I think I went overboard on potential failures with the Kyro2 and if heat is not the issue then the Kyro2 chips if quality made should last just as long as any other video chip.

Awesome about the mobile chips. T&L included you don't say. Why would a T&L unit be included there unless it was important heh?? Still this maybe where the Kyro really takes off.

Edited by noko on 04/27/01 11:17 PM.

noko · Apr 28, 2001

Updated above post to the source of benchmarks.

I did see the pattern and mentioned that to you today. I am looking to the second part of the review at Ace's Hardware.

Either way as long as the benchmarks are accurate even if done manually with a stopwatch it is still valid for comparisons. If not then your 10FPS increase you said you noted is worthless.

Guest · Apr 28, 2001

A stop watch couldn't be used in Giants. The frame rate counter refreshes very fast and its impossible to read individual framrates fast enough to get an average. What I did when I tested it to see what the features would do to framrate was stand in one place with those features off and take the lowest framerate I could see. Then goto the menu and turn the features on without moving, then I took the lowerest framerate that I could see again and it was 10fps lower, when I tested with DX7 the framerate didn't change when adding those features. What I did was more valid for comparisons then what you'd have to do with numerous cards because with lots of cards you'd have to keep restarting the game and making sure your in exactly the same place when taking the sample framerate. It is still ok for comparisons I was merely pointing this out to you incase you didn't know that there is no acual bench for Giants.

<<<<<So the boards being tested are not the ones that will be sold to the public? This is getting somewhat confusing. I thought the Hercules boards where pretty much the final boards that are going to be shipped.>>>>>

It really depends which rev board was used for the Herc review boards. The boards are final now but the review samples certainly aren't final, as I said there only 4 layers and the final board is 6 layers.

<<<<<Awesome about the mobile chips. T&L included you don't say. Why would a T&L unit be included there unless it was important heh?? Still this maybe where the Kyro really takes off.>>>>>

Another way of looking at it is that when IMGTEC have thought its important to include HW T&L they've done it without a problem in PowerVRMBX so they can do it if they see fit, why didn't they include HW T&L in Kyro II?, they had there reasons obviously but those reasons didn't include a lack of HW T&L tech.

noko · Apr 28, 2001

Could you explain the rather disappointing Anisotropic filtering degradation that occured on the Kyro2 at Ace's Hardware review?
<A HREF="http://www.aceshardware.com/Spades/read.php?article_id=25000232" target="_new">http://www.aceshardware.com/Spades/read.php?article_id=25000232</A>

Also any idea if redoing the board from 4 layers to 6 layers could have a negative effect on the cards stability. Obviously it will allow for a smaller card but how about stability? If it doesn't affect stability how come the 6 layer boards where not ready for reviews?

Edited by noko on 04/27/01 11:08 PM.

Guest · Apr 28, 2001

at least in trilinear/bilinear is better !
but you are forgeting something that is really interesting...

with the new driver from stm kyro can force texture compression in 99% of all games...

so kyro owners will get allways this feature on ...

forcing texture compression will give a boost on kyro 2 on anisotropic filtering (and trilinear)
I read in some place that it can get more than 50-100 % increase in performance...
but I am not sure...

not forgeting that the quality gain is not that great against trilinear...Edited by powervr2 on 04/27/01 11:49 PM.

Guest · Apr 28, 2001

Just been looking at that review you pointed me too (just a quick look so far) and as far as I can see the results on that page are asstounding. When you mentioned the Kyro II's bad performance with anisotropic filtering on the grand prix benchmark and also the fall in performance from the bilinear I assumed the Kyro II had done badely in grand prix, but nope the Kyro II roasts every card there with bilinear filtering, even the Geforce 3 is left behind by the Kyro II, it beats the Geforce 3 by 5fps, the Radeon by 50fps and the GTS by a stagerring 60fps!!!! the MX is beaten by the Kyro II by 70fps which is just totally embarasing infact all cards are embarased by the Kyro II in that test and at 1024x768x32 too, a very playable res. Also even though it takes a bigger hit for trilinear then the other cards it still beats every card but the Geforce 3 in trilinear filtering. Beating the Radeon by 20fps, the GTS by 30fps and the MX by (again and embarasing 40fps), it thrashes the GTS and Radeon until anisotropic filtering is used, one feature is not the be all and end all. I'd imagine the reason for the bigger fall in performance with trilinear and anisotropic is the real hit that ever card takes being shown up in a 100% efficient card, all the other cards are bandwidth bound so they have fillrate spare for better filtering without looking as if they need to use much more power so I suppose when you think about it what you should be mentioning is why are the other cards so slow with lower filtering. The Kyro II has 16 tap anisotropic and needs 2 clocks to do this, most benches I've seen with this sort of comparison show the Kyro II dropping its framerate in half with anisotropic (because of 2 cycles per pixel) but these benches show a much more extreme hit, I don't know what to tell you there really, I can't explain such a big hit, never the less unless your obsessed by anisotropic filtering then the Kyro II is awsome in that bench, infact awesome doesn't even come close to describing its performance in that bench.

Guest · Apr 28, 2001

there are already some final boards out there please noko wait 1 week more. They will try to overclock it !!!

" Well, after some trouble today Hercules/Guillemot has sent two boards to my testlab. The official retail version of the Kyro II and a Geforce III board. Guess that will mean for me to close the doors of my appartment and go into test mode. The extra good news is that I have about 3 systems to benchmark on. Altough I think I will bring that down to 2. The most important part of these machines is that one of them is based on a 1 Ghz AMD Athlon backed up with 256MB of 2100DDRAM!!!
On top of that we have some really great surprises for our readers. With really great I mean really really exciting stuff. Watch this site...next week"

<A HREF="http://www.beyond3d.com" target="_new">http://www.beyond3d.com</A>

why a final board will be better ?
the test samples are 4 layer boards while the final product will be 6 layer. More layers generally leads to more stability, less noisy signals and thus more overclocking potential.

Edited by powervr2 on 04/28/01 00:15 AM.

noko · Apr 28, 2001

Anisotropic filtering is very important and Bilinear filtering is rather crude. Here is a link showing the difference with and without
<A HREF="http://www.ati.com/na/pages/resource_centre/dev_rel/sdk/RadeonSDK/Html/Samples/OpenGL/RadeonAnisoFilter.html" target="_new">http://www.ati.com/na/pages/resource_centre/dev_rel/sdk/RadeonSDK/Html/Samples/OpenGL/RadeonAnisoFilter.html</A>

Beyond 3d has this to say about Anisotropic Filtering:

Anisotropic filtering:
Conventional texture filtering techniques do not compensate for anisotropy--the elongation of the screen pixel when it is mapped into texture space. This results in either blurring or aliasing, depending on the choice of texture level-of-detail. To achieve sharp textures, a card can use anisotropic filtering, a process that involves an elliptical kernel whose shape and orientation depends on the projection of the destination pixel onto the texture map.
For a full overview on how Anisotropic filtering works, check here
<A HREF="http://www.ping.be/powervr/Anisotropic.htm" target="_new">http://www.ping.be/powervr/Anisotropic.htm</A>

Looks like the Kyro2 is deficient in giving good performance when doing anisotropic filtering a quality important feature. I always use Anisotropic filtering due to the dramatic effects it has on game quality, on the Radeon it has minimal effect on FPS. The Radeon in the pixel pipeline can do 3 filtered textures per pipe per pass. Kyro2 waste time doing Anisotropic filtering in two clocks per texture when the radeon can do 3 Anisotropic filtered textures per clock cycle in each pipe line. Meaning for the Kyro to do 3 Anisotropic filtered textures it will need 3 clock cycles for the textures and 3 clock cycles for the Anisotropic filtering, 6 clocks for just one pixel where as the Radeon does it in one. Pretty lame on the Kyro2 part, lame indeed. Thats why the Kyro2 takes a nose dive in Anisotropic filtering and the Radeon doesn't, just think what would happen if you had 6 textures on the Kyro2. It has nothing to do with bandwidth on the Radeon, you could almost say the Radeon superior filtering is for free.

I wouldn't brag to much about great framerates using Bi and having terible framerates using Anisotropic filtering which is by far the most important for having quality rendering.

Edited by noko on 04/28/01 00:22 AM.

Guest · Apr 28, 2001

I'd have to agree with noko. Bilinear is so old now, and quite ugly. I was totaly shocked how the kyroII fell so much and I can say that I probably won't buy that card because of it but hey, some people might like playing games with thick blur planes.

Of course with a card that has near 100% efficiency you'd really notice the differences in filtering methods but the fall the kyroII takes is just a design flaw. Edited by m_kelder on 04/28/01 00:41 AM.

noko · Apr 28, 2001

Hey buddy how is the upgrading going?

Guest · Apr 28, 2001

Still holding...
everything crazy right now, vid card market unstable, cpu's being cut right and left and the more I wait the closer dual athlons come so I keep fighting myself to buy soon or wait for the duals.

Decisions, Decisions

noko · Apr 28, 2001

Thanks for the link for the upcoming review of actual shipping boards for both the GF3 and the Kyro2. This should clear up some issues that I have. Hate having to wait a week though.

noko · Apr 28, 2001

Yeap, especially when AMD just announced another price cut and the nVidia cards are rapidily falling in price. I guess it will come to a point where you just can't wait any longer and will have to buy. Your future career and starting down that path needs to start or continue as soon as possible. One more note, the updated bios for my KK266 to fix the via south bridge file corruption problem killed my hard drive transfer rates from 21mbytes/sec to 9mbytes/sec, ouch!! Maybe a 760 chipset board using the AMD south bridge would be better. DDR ram is cheap now from Crucial. Hopefully I will figure out a way to increase my hard drive performance again. I rather have the old Via South bridge with ATA 66 vice the handicapped ATA100 crap.

Guest · Apr 28, 2001

The Kyro II still walks all over the Radeon and GTS with trilinear filtering though, the difference between trilinear and 16tap anisotropic is not even noticable most of the time IMO. I have to look at zoomed in pictures to really see much difference between them so AFAIC its an unimportant feature to me and I'm sure thats true of most people. I freely admit that if for some reason anisotropic filtering is really that important to you then don't get the Kyro II. But if it isn't then the Kyro II is an awesome card and you can't burry all that just because of slow anisotropic filtering which is what you seem to want to do.

I find it strange that you did nothing but complain about the anisotropic scores in grand prix without mentioning the majority of the results massively favoured the Kyro II over the GTS and Radeon and then said:

"I really don't see the Kyro2 clearly beating the Radeon or GTS 2. Where advance games are soon to be hitting the shelves the GTS and Radeon will look even better."

So you didn't see the Kyro II beating them both in grand prix?, or is anisotropic all that matters?. The Kyro II beat the Radeon in 4 bechmarks and the Radeon beat the Kyro II in 2 bechmarks that too me shows the Kyro II clearly beating the Radeon, the GTS and Kyro II are half and half neither is a clear winner out of those 2.

<<<<I'd have to agree with noko. Bilinear is so old now, and quite ugly. I was totaly shocked how the kyroII fell so much and I can say that I probably won't buy that card because of it but hey, some people might like playing games with thick blur planes.>>>>>>

Who's talking specifically about bilinear here?, it seems a few people want to talk about the Kyro II winning with bilinear filtering so they can stress the large difference between bilinear and anisotropic and declare the Kyro II's performance as useless but what you seem to be leaving out is that the Kyro II still kicked the GTS and Radeons asses with trilinear too, it beat both of those cards, when using trilinear, by a long way.

noko · Apr 28, 2001

If you are only looking at FPS and not the quality that goes with it then yes. Since you have a Radeon put it back in and play QuakeIII with Anisotropic filtering turned on to the max. Play with it and then turn it off. You will know what I am talking about. Same with Serious Sam. Tri is nowhere near Anisotropic filtering and I can sure tell the difference. You think that anyone can tell the difference between 70FPS and 120FPS? or 90FPS and 70FPS but I tell you what I sure can tell the difference between Tri and Anisotropic Filtering especially when the Radeon is still doing 70FPS and the Kyro doing a jumpy 20FPS. This will affect all games on the Kyro2. A serious flaw as far as I see it. Still if it doesn't matter to you or anybody else then why even put this feature onto the card in the first place? Why waste the time, money and the effort? BEcause it is a significant advancement in applying textures to an object thats why. M_Kelder is 100% correct that it is a design flaw. For flight simulator buffs this causes the mosiac patterns to shimmer and glide all over the screen where the Radeon has a smooth still back ground. Once again is the Kyro2 sacraficing quality for speed that is beyond what is significant? Now I want to see benchmarks using Anisotropic filtering in Serious Sam with the Kyro2 and the Radeon, I guess the Radeon would win there too.

noko · Apr 28, 2001

You tell me if you can't tell the difference between these two images:

<A HREF="http://www.angelfire.com/hiphop/fdtv/de_dust0001.gif" target="_new">http://www.angelfire.com/hiphop/fdtv/de_dust0001.gif</A> -no anisotrphy
<A HREF="http://www.angelfire.com/hiphop/fdtv/de_dust0000.gif" target="_new">http://www.angelfire.com/hiphop/fdtv/de_dust0000.gif</A> -128x anisotrophy

Also some QuakeIII scenes🙁notice the FPS on the pages)

<A HREF="http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_1.jpg" target="_new">http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_1.jpg</A>

<A HREF="http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_2.jpg" target="_new">http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_2.jpg</A>

<A HREF="http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_3.jpg" target="_new">http://webpages.acs.ttu.edu/cwuthric/Anisotropic_128_3.jpg</A>

That is why so many people fall in love with the Radeon, the image quality is just awesome and the FPS is more than sufficient. That is how I use my Radeon with max quality settings and it flys not like what the Kyro2 would do if I was to use Anisotropic filtering. I would have to compromise and use Tri filtering only. I rather have the Anisotropic and ditch the Kyro2.

Also remember these are jpeg images which degrades the real image quality of the Radeon, in other words the real image from the Radeon will just blow you away. That is if you know how to turn on the features, some people don't.Edited by noko on 04/28/01 02:59 AM.

Guest · Apr 28, 2001

noko there will be always some negative aspect on everything, this is not a perfect world!
😉
but take note on the conclusion of the reviewer:

"Fifty percent higher framerates than the GeForce 2 GTS (trilinear filtering, environmental bumpmapping enabled)! Absolutely amazing! "

more on that review:
"It seems that Formula 1 makes good use of texture compression (DXT-1), and also applies many textures (bump mapping, shadow map.. ) "
(shadow map ????, with that directx8 bug ??? )
maybe it's because of this that we don't see half frame rate on kyro 2 from bilinear to anisotropic. I am sure that in most cases with texture compression forced anisotropic performance will get close to half the frame rate of bilinear!
let see 120/2=60 fps ...
more than enough !!
Edited by powervr2 on 04/28/01 07:54 AM.

Guest · Apr 28, 2001

(noko there is a problem with those link, images)

yes I think that radeon is better than kyro 2 in anisotropic !
but if we force texture compression the performance of kyro even with anisotropic will still be high (I am not sure if it will beat radeon though )
In 99% of games we can force texture compression!

by the way if quality is a issue then Why not talk about FSAA ?
😉

comparing trilinear against anisotropic is like comparing coca-cola against pepsi... if you hide the label you will not be able to distinguish the flavour of the two...
😉Edited by powervr2 on 04/28/01 09:43 AM.

Guest · Apr 28, 2001

Noko,
Good work with the anisotropy, and it made me think of something else. You own a Radeon and (I think) I have seen you mention several times about the good quality DVD playback it gives, specifically better than the GeForce line. I do not know a lot about what gives good DVD playback and how a graphics card effects it, so I wanted to ask you about the specs from the GeForce 3:

High-quality HDTV/DVD playback, as follows... 
*High-definition video processor (HDVP) for full-screen, full-frame video playback of HDTV and DVD content
*Independent hardware color controls for video overlay
*Hardware color-space conversion (YUV 4:2:2 and 4:2:0)
*Motion compensation
*5-tap horizontal by 3-tap vertical filtering
*8:1 up/down scaling
*Per-pixel color keying
*Multiple video windows supported for CSC and filtering
*DVD sub-picture alpha-blended compositing

Obviously this is no substitute for actually seeing it in action. I was mostly wondering how those specs compare to the GeForce 2 and Radeon. Does it look like they have improved DVD playback quality at all on the GeForce 3? Also, I haven't seen any reviews that really dig into this aspect of the GeForce 3 so if you know of any send them my way. 😎

By the way, the specs were taken from <A HREF="http://www.nvidia.com/Pages.nsf/Lookup/GeForce3_Product_Overview/$file/GeForce3_Product_Overview.pdf" target="_new">this</A> NVIDIA PDF file.

Thanks,
Warden

Guest · Apr 28, 2001

Do you think that kyro 2 and radeon 16 tap anisotropy is better than the 8 tap anisotropy of geforce 2 ?
maybe it's!
but I think that you are unable to tell from each other, it's like trilinear filtering vs anisotropy...
very dificult to distinguish from each other...

I noticed somewhere that with trilinear and anisotropic filtering the volume of textures passed increases a lot over bilinear filtering (maybe this is some bandwdith issue with that SDR ram of kyro, I don't know ).

So with texture compression we will get a big boost on the performance of kyro 2 (with trilinear and anisotropic).
if it wastes 2 cycle for doing a pixel with anisotropic then half of fps from bilinear...
when the bandwidth is not limited...
I think the feature of kyro that provide texture compression in any game (even those without it) will be great for kyro 16 tap anisotropic...Edited by powervr2 on 04/28/01 10:49 AM.

Guest · Apr 28, 2001

see the difference on trilinear filtering and anisotropic filtering:

first example:
<A HREF="http://www.ixbt.com/video/images/kyro2/kyro2-tlf2.jpg" target="_new">http://www.ixbt.com/video/images/kyro2/kyro2-tlf2.jpg</A>
<A HREF="http://www.ixbt.com/video/images/kyro2/kyro2-anis2.jpg" target="_new">http://www.ixbt.com/video/images/kyro2/kyro2-anis2.jpg</A>

Second example
<A HREF="http://www.ixbt.com/video/images/kyro2/kyro2-tlf3.jpg" target="_new">http://www.ixbt.com/video/images/kyro2/kyro2-tlf3.jpg</A>
<A HREF="http://www.ixbt.com/video/images/kyro2/kyro2-anis3.jpg" target="_new">http://www.ixbt.com/video/images/kyro2/kyro2-anis3.jpg</A>

booth shots with kyro 2

yap anisotropic is better ... and we can notice the difference with trilinear, but we are only able to distinguish from each other on these worst case scenarios
(mosaic pattern at distance).
A flight simulator landscape with mosaics are not good things I think...

I have a question to Teasy:
will it be possible to use 8 tap anisotropic (like geforce 2) thus giving less performance penalty ?Edited by powervr2 on 04/28/01 10:19 AM.

noko · Apr 28, 2001

The second example shows it better than the first series but also look at the textures on the walls and you can see that the more distant textures are significantly improved using Anisotroic texturing. Then again this is a individual perference and some people will think this is all silly while others will find it more serious. Yes we are prying open and looking at some of Kyro2's weaknesses but also its strengths. I believe a informed individual will make a better choice in buying something if told the straight scoop. Eventually we will get to a break down of the strengths and weaknesses of the Kyro2 and if those problems areas are fixable. Still a decent chip overall and will perform well for many people. The above links on my posts still work for me so not sure why they don't work for you. Oh yes I can distinguish between Pepsi and Coke when the labels are hidden. Also dintisguish betweeen plastic bottle Coke, canned Coke and my favorite Bottle Coke.

Edited by noko on 04/28/01 12:11 PM.

noko · Apr 28, 2001

Here is the white paper in PDF file on the Radeon Digital Video which is somewhat extensive. Allot of your questions can be answered there. The Radeon video side is very advance and it is like comparing the GF3 DX8 abilities to the Kyro2 DX8 abilities. Please read over it and when I get a chance I will compare the difference between the GF3 which apparently didn't add anything extra then what the GF2 line of chips had. Please note you will have to have Adobe Acrobat Reader installed to view.

<A HREF="http://www.ati.com/na/pages/resource_centre/dev_rel/atirdv.pdf" target="_new">http://www.ati.com/na/pages/resource_centre/dev_rel/atirdv.pdf</A>

Edited by noko on 04/28/01 01:50 PM.

Kyro 2 - Part 2

Guest

Guest

Guest

Guest

Distinguished

Distinguished

Guest

Guest

Distinguished

Guest

Guest

Guest

Guest

Guest

Guest

Distinguished

Guest

Guest

Distinguished

Guest

Guest

Distinguished

Distinguished

Guest

Guest

Distinguished

Distinguished

Guest

Guest

Guest

Guest

Guest

Guest

Guest

Guest

Guest

Guest

Distinguished

Distinguished

Share this page