A PS3 breakdown of all components.

[EbonySeraphim]

Acording to the author of the article the PS3 walks all over the 360 with out breaking a sweat. If one peice of hardware is so vastly superior it should have no problems destroying the compitition graphically. If the author is right where are the results or are all developers stupid?

So the RSX has no extra sampling units or extra compression when MSAA is used like ATI GPUs? If it does the it has hardware MSAA.

Please don't ever say something that ignorant about my post/article. Additionally, if you want to argue the merits of the game console superiority over one another using games as comparison bar, then there is another section in this forum, or other forums for that out there. I'm pretty sure "Console Technology" isn't supposed to be much of a vs thread where people toss around console "bests" with subjective opinion.

How can you make it better without pointing out what's wrong? Your entire statement, while sounding reasonable in theory, has not stood up in practice in any of the discussions I have seen. The problem with bias, is that it even when logical empirical evidence is presented to refute another point, bias can easily discredit, ignore, or argue such findings. Particularly in areas such as this, where clearly there are no right or wrong answers, and passions run so deep. Admitting bias and a lack of understanding of the compared architectures doesn't help to bolster the validity of your article. And contrary to your belief, there are those who can remain objective, and accurately and fairly present the facts whether they are personally bias or not. None of us are perfect, and the criticism was not and is not a personal attack. Good luck with the article.

Bias only appears to refute logical and empirical evidence when the audience, or the participants have a weak understanding of what's going on. If you think you're losing to a person bringing illogical arguements to the table, you are more than likely feeling like the audience is being suckered into it. It's your job to present your arguments in such a way that your opponent cannot deny it's truth, and others cannot deny it too. If someone is throwing seemingly valid arguments that no one understands in a debate, they lose to whoever is bringing the easier to understand arguments to the table. Even if your arguement is the more technically deep, if you have high enuogh technical understanding, you should be able to share that knowledge down to the level of educating people on what they should know, and then explaining the arguement.

Pointing out the bias is not pointing out what is wrong unless the bias is actually linked to me arriving at an incorrect conclusion - and you're actually mentioning what the incorrect conlusion or facts are. Telling me "you're wrong because you are biased" is pretty redundant. Of course a person with bias will have the relatively wrong answer that is aligned with their bias. "Fixing it" means laying down what's factually, logically, and maybe even empirically wrong.

And contrary to your belief, no one is unbiased unless the knowledge is extremely basic and dependent only upon simple definitions. 2 + 2 = 4, is even a biased statement that assumes a number system people on this world are used to and assumes the behaviour of two mathematical operators. "Terrorism is bad" is a biased statement. The term terrorism is subjective in the sense that an organized government or counry may not be recognized by terrorists, and the term "bad" is subjective as a whole. Saying "developers can do this" is biased, because there are going to be developers who feel it would cost too much or don't know how to do it, and thus say "developers won't do this," ignoring the possibility altogether. Saying PS3 > PS1 is a biased statement given that I could name something that the PS1 has that the PS3 doesn't. Subjectively, I could scale that one little thing the PS3 doesn't have extremely high, and down scale the horsepower advantage of the PS3.

If any attempt is being made to compare two hardware architectures that aren't exactly the same and each one has something over the other, then the act of establishing which one is better is biased. Any article that tries too hard to not be biased when comparing things, ends up being one of those comparisons that make the two equal and pretty much worthless. A comparison like that should just have aimed to say they are different and there is no conclusion.

Did I say that admitting my bias makes the article any better? No, the post is already written and if it stays that way, it will be how it is. Admitting my bias only enables me to take a second look at it later, and identify where it may be easier. Otherwise, I am also asking for help, advice, information, and research to help be put into the article to have it be based on more factual premises and less biased ones. Again, if you only care to prove to me that my article is bad; that I am biased; or that there are people/sources less biased than me, your efforts are pointless as it's already known, and not helping anything.

But there is no learning curve. The unified shading is totally transparent to the developer. It's DX9 all the way. The problem is the EDRAM which is currently more of a bottleneck than a feature for most developers.

And this is just something the PS3 has not to struggle with.

There is no learning curve if you want to accept ATI's algorithm for dynamically scheduling the shader units. From that angle, DirectX is probably used directly, but I'm pretty sure that the load balancing can be controlled for optimization if ATI's algorithm doesn't suit a game well. ATI is relatively confident that most games will run well on what they have in place already (at least for a while).

 

[Tap In]

this has gone from a (potentially interesting) technical discussion to an off topic philosophy of bias.

 

[Acert93]

There is a difference between bias, having an agenda, and drawing conclusions BEFORE taking the time to fairly evaluate the entire body of information. These are all degrees and not remotely equivalent.

Obviously we are all prone to degrees of ignorance as none of us are experts on all things [and most here have never touched a dev kit, so our knowledge is at the mercy of those who seek information from] and make mistakes, but we should be careful--especially when claiming a degree of experience, knowledge, and specialty in a general area--to do due diligence in the research and analysis of our project before making public conclusions. Being knowledgable means more responsibility.

My bias further comes in play when I see that one of my goals (which I wasn't concious of until reading early comments) was to show that the PS3 brings benefits to the table, and that it is possible to develop for. It's pro-Sony due to my ignorance of the Xenos

Disregarding agendas, drawing objective opinions requires a broad evaluation of all relevant material on a subject. An example: I surely cannot talk about global warming if I have failed to collect meaningful information about 50% of the globe. Any analysis lacking the this information will be lacking relevant contextual data and the conclusions drawn from such would be harshly criticized for violating basic principles of investigative research. As they say, "Get your ducks in a row".

As you will continue to learn in college that there is a proper method of investigation. Drawing conclusions without thoroughly evaluating the topic will not only lead to faulty and misleading conclusions, but it will also dent your credibility because you are willing to arrive at conclusions without objectively surveying all the relevant material.

In lay terms we would call this self fullfilling prophecy.

Although I am biased, invalidating the entire post's content is pretty stupid.

No one has said your post is completely invalid. Quite the opposite. Further, most of the information in your posts are actually on this very forum and have been so for over a year. You will find that a large portion of your facts are true and agreed upon on this forum.

But there is a fine line between facts and conclusions being drawn from those facts. Its the method and slant that you will find that certain posters take issue with.

Even textbooks are biased.

See above. If you cannot see the difference between objective bias and agenda driven bias there is no point. Any textbook worth its salt is going to have a) opinions tempered by the consensus of the academic community as a whole and b) where a decision must be made to present one side over another noting that there is an adverserial side co-competing within the paradigm, thu not to illinform or mislead the reader.

Further, it is pretty standard in academia [where a textbook would be created] that you recognize your bias and then temper it with informed opinion that engages the entire spectrum of facts and explaining why you disagree with other positions--knowledgably and knowingling--without distorting or misunderstanding the topic.

Comparing forum posts and poster bias to a textbook, as if such bias is even within the same ballpark, is silly at best.

There is no learning curve if you want to accept ATI's algorithm for dynamically scheduling the shader units.

The context of this point by the original poster was not just the shader units, but the eDRAM as well.

And yes, there is a learning curve with using Xenos. Compared with the dev kits that MS developers used until Augest 2005 there are many significant differences. Xenos is a significant departure from the desktop GPUs in the PC marker, the GPUs in the dev kits (9800 and X800), and NV2A. Besides unified shading architecture [and API] and eDRAM there is a significant change in "pipeline" balance (3:1 math to texture balance versus 1:1 in previous generations), full SM3.0 support (32bit shader percision, dynamic branching in pixel shaders, vertex texturing), and so forth. This is in addition to featureset additions. Very few developers have touched SM3.0 and it is hardly a baseline at this point. Toss in having fewer ROPs (but more bandwidth) and you have something significantly different than what developers have been using.

Not to go off tangent, but this is one of the smart advantages (at least for the short term) Sony jumped on: traditional desktop GPU that had an architecture on the market in Spring/Summer 2004. And by partnering with Nvidia Sony was able to leverage SLI which allowed developers in early 2005 to get a rough handle on

a) featureset and feature performance
b) overall performance ballpark

If Sony had gone a route where they had to wait until 2006 to get RSX-like features and performance for developers to test and begin developing on Sony would have felt a significant crunch.

Instead they get a derivative of NV's Q4 2005 flagship GPU, built on an establish architecture from 2004, and a part that is not only well known but comes with great tools (Cg) and will allow excellent porting on the graphics side from the PC and PC derived engines.

Considering MS's early reliance on "portware" this strategy would have probably been better at launch. But by all accounts Xenos is more forward looking than hardware to port current PC titles. How well Xenos falls in line with MS's vision has yet to be determined. How you look at the first 8 months depends on a lot of variables.

 

[Acert93]

Well, in simplest terms, it's generally a pain in the neck if you try and treat the Xenos like the GS and start going through framebuffer-wide blend operations or try to accumulate using backbuffer copies.

But you would have to ask why anyone would even try to use Xenos like GS? Xenos eDRAM does not even have the same functionality of the GS; it is akin to trying to performance complex pixel shaders on GS. Xenos eDRAM was not designed to work like GS--but this does not invalidate the bandwidth it has for its intended uses. But your point on MRT is well worth noting and it will be something interesting to watch over the next couple years.

Anyhow, the above is an example of why such comparisons are fundamentally flawed: You can draft situations outside the realistic use of the hardware that can make 6 year old hardware superior in specific circumstances (which indeed they may). Yet being realistic GS was hardly superior to NV2A in terms of the pixels displayed on screen and it had 6.4GB/s of total bandwidth for GPU and CPU (which could use up to 1GB/s). So is all that bandwidth a net win?

New generation, new problems. This has been the mantra for Cell development since day 1. The only difference is the problems on the graphic end tend to be more well defined, but in turn sometimes the solutions are more confining. GS is a good example of this: they gave developers a boatload of bandwidth to the eDRAM (which was much more robust in accessiblity than Xenos eDRAM) but the GS lacked many, many features on the logic side of things. It solved some problems, yet left others with inadequate resolutions.

Of course your a dev, so I don't need to tell you that

 

[EbonySeraphim]

Acert93:
Now that we agree on what bias is and all of the different forms of it, and that I have exhibited bias making my post quality worse than journals, newspapers, and even the internet as a whole. Can you look beyond those flaws and actually give a specific correction or point of improvement? If you can't and aren't, then I'm done conversing here.

 

[Bigus Dickus]

1. Bandwidth. You poo-pooed earlier MS marketspeak that simply added all the busses together (22 + 256 or whatever), and then proceded to make your own similarly useless "comparison" of bandwidth by adding the GDDR and XDR bandwidth and comparing it to 360's GDDR bandwidth. A truly objective approach would have been to at least attempt to dissect the bus usage for both cases, including Xenos eDram usage. That very thing has been done quite extensively on this forum. As a crude summary, you should examine what things the RSX might actually use the XDR for, and how much bandwidth that frees up for GDDR usage. And on the Xenos side, examine the bandwidth savings of eDram under normal usage and how much that frees up for GDDR usage. You might then be able to begin making an objective comparison of bandwith. However, you started with an agenda, that MS claims were overhyped, and that PS3 had two busses, and your approach was to make a very simplistic comparison in the most favorable way for the PS3.

2. USA. I know that a lot has been said on this point, and that you have ammended your earlier post, but this is obvioulsy an area where you didn't do much research. Not only is ATi transitioning their PC products to an unified architecture, but so is nVidia most likely in the future (doubtful for the next generation with G80). USA isn't about raw performance so much as flexibility and efficiency. If all you had to do was texture samples, like the DX6/7 generation, it is easy to do that a hell of a lot more powerfully with less transitors than either Xenos or RSX. But USA is a forward looking architecture that not only allows for maximum pixel shading power for the fewest transistors, but also far more powerful vertex setup and shading than traditional architectures, and provides the opportunity for devs to tailor their games to do some interesting things.

3. A complete lack of discussion of what are well known architectural details of Xenos. Sure, you claimed your ignorance here up front, but in an article where you start by trumpeting your credentials and beating your chest, and otherwise playing the part of an author making an authoratative comparison of the two consoles, I think it is only reasonable that you do research on both architectures under comparison. Otherwise, you should have simply left off all mention of RSX and Xenos altogether, and presented the article/post as a comparison of CPU architectures. But having dipped your toes into the GPU bucket, you should have mentioned things like FP format HSR support, memexport, hardward tesselation, cache locking, etc. Some useful, some perhaps not, but certainly worthy of a mention in such a "comprehensive" comparison.

Those are just three areas upon an intial skim of the article that I found glaringly insufficient. It wouldn't be so bad, had you not presented the article in they way that you did. But on a PS3 fan forum, with a long preface about how knowledgable you are on the subject, and post after post afterwards proclaiming you 'one of the most console hardware knowledgable persons out there' etc., I think it is understandable the reactions some people have had to it.

Oh, and don't come here and cop an attitude. I hardly think after four posts you are an expert on which subforums are best suited for what subject matter, nor should your first visit here consist mostly of chastising long standing senior members for seeing through your thinly tech-veiled agenda. If you stick around, you might find that this forum contains all the information you needed for your "research" as well as some of the most knowledgable persons around. I'm hardly one of those, but then again I didn't write a "once and for all" analysis of the PS3 and 360 architectures.

 

[Geo]

Let's keep it substantive, please. It also isn't easy to come in here as a stranger.

Certainly the staff would like to encourage authors of pieces from around the web to show up here to talk about them. We would hope the members would as well.

 

[Shifty Geezer]

But you would have to ask why anyone would even try to use Xenos like GS?

This was because of my statement likening Xenos eDRAM to PS2's eDRAM, in saying that neither will be starved of BW for framebuffer effects. My fault for not differentiating different effects on different platforms. For XB360 I'm thinking the framebuffer filtering, testing, and so forth 'effects' and not any postprocessing style 'effects'.

I not sure how to more clearly phrase it. That PS2 was designed to have more than enough BW for it's tasks, and similarly Xenos is designed to have more than enough (or just enough for absolute peak throughput) BW for the tasks it is designed to do, though they're different tasks. ?

Anyway, the point is Xenos' eDRAM has a significant role to play in the system in elliminating any bandwidth concerns of a number of framebuffer activities, whereas on other systems (PS3 and standalone GPUs) those same activities could saturate available BW and starve the GPU.

 

[Jesus2006]

Anyway, the point is Xenos' eDRAM has a significant role to play in the system in elliminating any bandwidth concerns of a number of framebuffer activities, whereas on other systems (PS3 and standalone GPUs) those same activities could saturate available BW and starve the GPU.

Agree. But you must add here that this is not free, otherwise it's biased as well

It's not something you turn on and get a benefit from it. It has several (severe) flaws, like raising geometry, problems with framebuffer effects that overlap tiles and z-only pass rendering. And this is simply because the EDRAM is basically too small for it's main purposes.

EDRAM would have been perfect if everything fitted in it (i think the PS2 framebuffer entirely fitted in there, or had to ) but the way it is now it is some kind of bottleneck as well.

And still my question remains: Why are there PS3 games outputting 4xAA seemingly without much problems (theoretically starving from its lower framebuffer bandwidth) when the same thing is hardly found on the 360, despite its eDRAM?

 

[Jesus2006]

One thing to add:
For Xbox 360, i think it's not only a benefit, but it is absolutely necessary for the system to run at a decent speed. I guess without it it would be nearly impossible to run modern games on it while the GPU and CPU utilizing the same bus with heavy bandwidth demands, blocking each other.

 

[Shifty Geezer]

It's not something you turn on and get a benefit from it. It has several (severe) flaws, like raising geometry, problems with framebuffer effects that overlap tiles and z-only pass rendering. And this is simply because the EDRAM is basically too small for it's main purposes.

EDRAM would have been perfect if everything fitted in it

And was actually producable at that size! You seem to fail to understand that all these have to weigh performance, efficiency, and development ease with produceability and cost. MS could have designed the system with 24 MBs of eDRAM, but at what cost? Would such a chip even be manufacturable? Instead, they chose a system that balances the relative costs of eDRAM with the advantages it brings to the rest of the system, like not needing expensively fast main memory and split memory pools. Tiling is a way of balancing cost and performance, and the costs of tiling aren't as horrific as you seem to think. If MS had forsaken eDRAM because it wasn't a perfect solution, they would have had to go the way of PS3, which isn't a perfect solution. Or put in 24 MB of eDRAM for a perfect GPU and charge $1000 for the console, or lose $600 a unit in loss-leeding hardware...

Because there is no perfect solution, no consoles would ever get made! All these devices are a collection of compromises over different criteria. These technical evaluations should be a list of strengths and weaknesses, identifying the compromises made and what (dis)advantages they bring to the system, and outlining how developers will have to work with the system. Then leave it to the reader to decide which solution is better, which will always be subjective.

 

[Arwin]

I think a significant contribution on the PS3 is probably that the textures can be copied in from the XDR memory because the RSX has fast connections to both its own GDDR3 and the XDR memory. Thus the Cell could work on and prepare textures and vertex data, the RSX takes that from XDR and finishes the scene in GDDR memory. This seems to me to be a pretty efficient pipeline. At least, that's what I've understood from reading comments from developers so far ...

So we'll have to wait and see how and whether the 360 will solve this problem. The theory seems to be tiling, but how well is it going to work in practice? But it's a fair comment that developers have had precious little time to work this out so far, so I'll reserve judgment until January.

 

[Jesus2006]

Because there is no perfect solution, no consoles would ever get made! All these devices are a collection of compromises over different criteria. These technical evaluations should be a list of strengths and weaknesses, identifying the compromises made and what (dis)advantages they bring to the system, and outlining how developers will have to work with the system. Then leave it to the reader to decide which solution is better, which will always be subjective.

Agree, there is no perfect solution right now, but that does not mean that one isn't supperior over the other. Currently from the things we know and see it seems that the PS3 solution might be the leading* one, otherwise i cannot explain to me why it is as it is

*or at least the one that developers get most out of with less effort, which currently seems to be more than on Xenos, looking at the games/AA levels

 

[Laa-Yosh]

How about you and me both shut up for this thread because neither of us knows enough about the respective system's architecture compared to the senior members. We're just derailing the thread.

That is quite right.

 

[Shifty Geezer]

*or at least the one that developers get most out of with less effort, which currently seems to be more than on Xenos, looking at the games/AA levels

I haven't yet heard any developer say writing for PS3 is no effort! Actually, I remember one quote, some guy talking about writing shaders for RSX. But as a system, no-one's saying it's easy. You won't find any devs saying it has a better development environment than XB360 either! I'd like to hear how the AA is being produced first before deciding whether it's easy or not

 

[EbonySeraphim]

Bigus Dickus:
Don't skim. When you skim, you miss things. When you miss things, you claim I didn't say or explain something which is actually there. Don'tr say "you didn't do this" unless you have exhausted everything I wrote. Considerations for a lot of what you said are there, but aren't there to such an obvious degree because I didn't want to write a 200 page book, and I assumed people would have some knowledge and inferential ability.

First thing you should realize is that the primary purpose of the post was NOT to compare PS3 vs Xbox360 to the extremes. As much as I'm aware that this is what most people care about, it was secondary to trying to establish what the PS3 is in itself. So while I care about being more informed about the Xenos and other competing hardware to the PS3, research on the "others" was limited. I am a human being. I have time constrains and have to put a finite amount of effort and work into something before I decide to just push it out there. There will always be a level of incompleteness in the research done. Note the keyword limited and not non-existent - even if it seems as if I didn't research at all. After the fact, yes, I can see I lacked enough research so tack such a relatively sloppy comparison next to a pretty solid assessment of the Cell. But I was NOT talking out of my ass when I wrote the original post. I was fully aware that the eDRAM takes bandwidth off of the main memory bus of the Xenos. I was fully aware that the RSX using XDR RAM shares bandwidth from the Cell. My "agenda" was to counter the original claims, but not for the reader to simultaneously "forget" that the prior claims had some validity in it. If all you picked up was "PS3 has two buses" and "MS was overhyped" then you are oversimplifying what I wrote and should stop talking about it. If you did a little more than skim, you might have seen some pretty obvious indicators that show what you just tried to educate me about, was already understood and known.

I did not lay out my credentials to establish any sort of dominance on the matter. If you think I was being cocky in doing so, you are out of your mind. I'm well aware that there are people who've been in the industry more years than I've been alive, and those who have masters degrees and maybe even PhDs. Me putting my credentials out there serves as a Achilles heel more than anything else as you are well demonstrating. The reason it was put there, was to establish that I wasn't talking about of my ass to the best of my knowledge - and that knowledge is relevant/substantial, not "I know all, listen to me."

This stake that some people seem to bent on driving further through me, is making me "cop an attitude." I think some of you guys are still angry at my ignorance that I'm fully aware of, and you still keep on mentioning it without adding anything beneficial. How many people would learn happily on any subject when the teacher keeps telling them "you're stupid" before actually teaching what it is that needs to be taught? Take that another step worse, and have the ridicule come along with a weak or non-existent lesson? Any student (in college) would just leave the room because the teacher is clearly wasting their time if their only action is to call students stupid with minimal lecture.

Ah yes, and if you're so angry at the sloppy assessment in the post of "PS3 > XBox360" regardless of if there are counter-merits or not, then you are probably just angry because of your own bias and are being very hypocritical in slamming me about mine.

You also misunderstood "conversing here." I didn't say that on the whole scale of the Beyond3D forums. It was with respect to this thread. I am already doing added research on the topics I don't/didn't know about so you don't have to have to encourage me to stay or leave. If I find what I am looking for here, I'll get it. Also, no matter where I go, in a different MMO or a new forum, I don't really care who's a senior member or not. If they are that great, then they'll show for it in the content of their posts or what they do regardless of if they have 4 posts, or 2000 posts, been there for a few weeks or a few years. There are senior members of any playground that are still quite misinformed, irrational, or strongly biased. No one should demand respect, it should be given to all, and only removed when warranted.

Let's not forget that I posted my "comparison" on a forum as a post, not an article worthy of any sort of press release. It has somewhat risen to the level of an article given where it has been referenced. Of course there is reason/need to critique it on that level if you aren't on the PS3Forums, but you guys are out of line if you think I was trying to be authoritative on the level of what might be posted on IGN, 1up, or even Beyond3D.

...and post after post afterwards proclaiming you 'one of the most console hardware knowledgable persons out there' etc., I think it is understandable the reactions some people have had to it.

Thanks for throwing bullshit like this out there. Makes me feel like I am conversing with someone reasonable when something is brought up that is completely false.

To others: Sorry for posting some of this crap here. I am reading the other discussion that is beneficial to me on these forums and in this thread. I'm just hoping those who *might* actually have something helpful to say, get it out instead of merely critiquing what is wrong.

 

[Cheezdoodles]

A lot of people here had a lot of helpful tips to you. You just chose to ignore it because you have no clue about a lot of the things you have drawn conclusions about. Instead of actually discussing or being thankfull for the parts thats pointed out to you, you rather wanna go and attack some random guys that called you biased (which you are but thats not the point). Biased or not biased, don't draw conclusions your not knowledgeable enough to make!

Like some other people here pointed out, read up on unified shaders, amongst other things.

And what was that crazy part about PC GPU's being bottlenecked because of its PCI-E connection? I had a serious laugh over that one.

 

[EbonySeraphim]

A lot of people here had a lot of helpful tips to you. You just chose to ignore it because you have no clue about a lot of the things you have drawn conclusions about. Instead of actually discussing or being thankfull for the parts thats pointed out to you, you rather wanna go and attack some random guys that called you biased (which you are but thats not the point). Biased or not biased, don't draw conclusions your not knowledgeable enough to make!

Like some other people here pointed out, read up on unified shaders, amongst other things.

And what was that crazy part about PC GPU's being bottlenecked because of its PCI-E connection? I had a serious laugh over that one.

No, you just think that is the case. I have been taking in the helpful tips and corrections people have been posting. Sorry for not being explicit about that and handing out thanks to those who have been helpful Other random people, unfortunately have taken up more space in this thread. It's not hard to figure out why.

I hope I didn't suggest that PCI-E was the bottleneck of games on PC GPUs. I suggested that GPU bandwidth to it's own video memory is plenty for games, and PCI-E isn't a bottleneck unless they are using significant amounts of system RAM or video RAM for bandwidth intensive operations. Of course, the bus is plenty fast if it is primarily being used for it loading and swapping textures in an out. I'm 99% sure that frame buffer effects will not go over this bus, and less sure that substantial texturing can go over this bus. Am I still wrong about that? Go ahead, and say it. There is no need to call the part crazy. I am reviewing the post and will look to see if that part wasn't worded properly.

 

[Cheezdoodles]

I hope I didn't suggest that PCI-E was the bottleneck of games on PC GPUs. .

In the version of the article i read, you did suggest that PCI-E was a bottleneck.

You even went as far as comparing it to Flex I\O and saying how the PS3 will be superior to PC GPU's because of it.

 

[dantruon]

from what i gathered so far on many forums including this one i come to the conclusion that the Xenos beat the RSX hands down in theory but in practice the RSX whipped the Xenos's ass painfully.