Digital Foundry ha tenido la oportunidad de conversar con el arquitecto de sistemas Mark Cerny y de ver a detalle el diseño y especificaciones técnicas de la nueva consola de Sony, la PlayStation 5. Acá en The Couch hemos compactado esa información para que conozcás un poco más acerca de detalles como la GPU, la SSD, el almacenamiento expandible y el Audio 3D.

La PS5 incorporará 10.28 TFLOPS, 16 GB de memoria RAM y un SSD, lo que proporcionará mayor rapidez en los tiempos de carga. También utiliza «sombreadores primitivos», similares a los sombreadores de malla que posee Nvidia Turing y las próximas GPU RDNA 2, esto ayuda a una optimización en la ingeniería geométrica. La GPU AMD RDNA 2 presenta 36 unidades de cómputo que se ejecutan a frecuencias limitadas a 2.23 GHz; en cuanto al conteo de transistores, la matriz de 36 CU de PlayStation 5 equivale a 58 CU de PlayStation 4, es decir, un 62% más efectivo.

Esta consola usa tecnología de CPU Zen 2 de AMD con comunicaciones previas que confirman ocho núcleos físicos y 16 subprocesos, la rapidez con la que se sincronizan las frecuencias es de hasta 3.5GHz; sin embargo, Cerny explica que depende de ciertas condiciones, esto podría ralentizarse hasta un 10%.

Una de las características más llamativas es el impulso, lo que permite que el rendimiento de la consola siempre sea el óptimo sin importar la temperatura ambiente. Según Sony, todas las consolas PS5 procesan las mismas cargas de trabajo con el mismo nivel de rendimiento en cualquier entorno. Cerny también explica que, en lugar de medirse la temperatura real de la matriz de silicio, se toma en cuenta las actividades que realizan la GPU y la CPU.

El rendimiento es notablemente diferente, porque los ‘teraflops’ se definen como la capacidad computacional del vector ALU. Esa es solo una parte de la GPU, hay muchas otras unidades, y esas otras unidades funcionan más rápido cuando la frecuencia de la GPU es más alta . Con una frecuencia 33 por ciento más alta, la rasterización va un 33 por ciento más rápido, el procesamiento del búfer de comandos va mucho más rápido, los cachés L1 y L2 tienen ese ancho de banda mucho más alto, y así sucesivamente.

Mark Cerny

El trazado de rayos funcionará acelerado por el propio hardware, no por software. Esto quiere decir que, por ejemplo, un juego podrá utilizar con éxito reflejos basados en trazado de rayos principalmente en complejas escenas animadas, todo sin tener que sacrificar rendimiento en la consola.  

Sony está apostando fuertemente a la calidad del sonido por medio de una unidad de audio 3D que use apenas una fracción de un núcleo Jaguar para ofrecer Surround 7.1. El nuevo Tempest Engine podrá soportar muchísimas fuentes de sonido a una calidad bastante superior en comparación a la competencia. Con esta ingeniería, se genera la sensación de estar realmente en medio del ambiente como nunca antes, especialmente con unos buenos audífonos.

Las unidades flash de PS5 conectarán los módulos de la memoria SSD a través de una interfaz de doce canales con un rendimiento de 5.5 GB/s con un almacenamiento total de 825 GB de espacio. Esto permite que 2 GB de datos se carguen ahora en una cuarta parte de segundo y los 16 GB de memoria RAM carguen en tan solo un par de segundos (sin exagerar). 

Por último, con respecto al almacenamiento expandible, Cerny admite que tenía muchas dudas al respecto, pero que ahora está claro lo que se puede hacer cuando se completen los 825 GB de espacio. PlayStation 5 es compatible con versiones anteriores y puede ahorrar espacio ejecutando sus juegos antiguos desde el almacenamiento externo estándar, pero también está la opción de que el usuario adquiera piezas de tecnología PC NVMe y las instalen en la consola; esto es un compromiso que tiene Sony en incentivar la adición de unidades que puedan optimizar el funcionamiento de PlayStation 5. En resumen, el almacenamiento expandible es posible y no necesitará unidades de Sony obligatoriamente.

Acá te dejamos un cuadro comparativo entre la nueva generación y la generación anterior de PlayStation.

Características PlayStation 5 Playstation 4
UPC8 núcleos Zen 2 a 3.5 GHz (frecuencia variable)8 núcleos Jaguar a 1.6GHz
GPU10.28 TFLOPs, 36 CUs a 2.23GHz (frecuencia variable)1.84 TFLOPs, 18 CUs a 800MHz
Arquitectura de GPURDNA personalizado 2GCN personalizado
Interfaz de memoria16 GB GDDR6 / 256 bits8 GB GDDR5 / 256 bits
ancho de banda de memoria448GB / s176GB / s
Almacenamiento internoSSD personalizado de 825 GBHDD de 500 GB
Rendimiento de IO5,5 GB / s (sin formato), típico 8-9 GB / s (comprimido)Aproximadamente 50-100 MB / s (depende de la ubicación de los datos en el disco duro)
Almacenamiento expandibleRanura SSD NVMeHDD interno reemplazable
Almacenamiento externoSoporte USB HDDSoporte USB HDD
Unidad ópticaUnidad de Blu-ray 4K UHDUnidad de Blu-ray

197 COMENTARIOS

  1. лицензия на банковскую деятельность, банковская лицензия казахстан мысық аттары
    қыздарға, мысыкка аттар ритуальные услуги цены, мусульманские
    ритуальные услуги алматы как
    подключить раздавай интернет алтел, алтел тарифы 590 тенге как подключить

  2. салыстыру әдісі балабақшада, заттар тобын салыстыру және санау балабақшада ондық бөлшек дегеніміз не,
    ондық бөлшек туралы мәлімет тусинде шириген жумыртка корсе не болады,
    мусылманша тус жору жумыртка
    панцерия противопоказания, бальзам короткова панцерия отзывы

  3. полушубок соболь цена, соболь шуба алматы формативное оценивание служит для, суммативное оценивание за 1 четверть
    8 класс power bank usb c lightning, power bank 20000 mah алматы музыка барум, nokian hakkapeliitta 10p suv 225
    60 r18

  4. This aligns with the «Signal Noise» theory we’ve been developing. You need enough noise to mask the signal, but not so much that you lose authority. delicate balance.

  5. Question: Have you tested this approach with expired domains? We’re running some experiments now and the results are… mixed. Your methodology seems safer.

  6. This is the missing piece of the puzzle for us. We had the content and the technical SEO, but the off-page signal diversity was lacking. Thanks for the clarity.

  7. I’d love to see a follow-up post on how this integrates with social signals. We feel there’s a multiplier effect there that isn’t being fully utilized.

  8. The shift towards «entity-based» indexing is real. Your strategy seems to leverage that by building entity associations rather than just keyword matches. Smart.

  9. One minor correction: the update rollout was actually 14 days, not 10. But that doesn’t change your main point—the volatility window is getting wider.

  10. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  11. This is exactly why we moved away from automated PBNs. The risk/reward ratio just doesn’t make sense anymore compared to what you’re describing.

  12. I’ve been following this topic for a while, and your analysis on the structural shifts really adds a new perspective. We’ve noticed similar patterns in our internal data at SignalLayer, specifically regarding the volatility timeline.

  13. I’ve been following this topic for a while, and your analysis on the structural shifts really adds a new perspective. We’ve noticed similar patterns in our internal data at SignalLayer, specifically regarding the volatility timeline.

  14. We’ve been A/B testing this exact hypothesis. Group A (your method) is outperforming Group B by 40% in terms of ranking stability. The data speaks for itself.

  15. Actually, I have to disagree slightly with the second point. In our testing, we found that over-optimization was less of a factor than pure engagement metrics. It’s interesting to see how different niches react differently.

  16. The analogy of the «immune system» is perfect. You need to build resistance before the virus (update) hits. Too many people react instead of prepare.

  17. Brilliant articulation of the problem. The industry has been too focused on metrics like DA/DR instead of actual traffic flow and user behavior.

  18. One minor correction: the update rollout was actually 14 days, not 10. But that doesn’t change your main point—the volatility window is getting wider.

  19. I’m sharing this with our content team. We’ve been struggling to explain why «quality over quantity» isn’t just a cliché, and this illustrates it perfectly.

  20. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  21. Have you considered the impact of mobile-first indexing on these placements? We’ve noticed that some «desktop-safe» strategies are flagging on mobile crawls.

  22. This is a solid breakdown. One thing I’d add is that the impact of these updates often lags by 2-3 weeks. We tracked this across multiple projects and found the recovery phase is where most people give up too early.

  23. I’d love to see a follow-up post on how this integrates with social signals. We feel there’s a multiplier effect there that isn’t being fully utilized.

  24. This complements the «Entropy» theory perfectly. If you don’t introduce randomness, you’re just painting a target on your back. Glad to see others advocating for smarter engineering.

  25. Just wanted to say thanks for the detailed case study. It’s rare to see actual data backing up these claims. We’ll be adjusting our Q4 roadmap based on some of these insights.

  26. This is the missing piece of the puzzle for us. We had the content and the technical SEO, but the off-page signal diversity was lacking. Thanks for the clarity.

  27. This is a solid breakdown. One thing I’d add is that the impact of these updates often lags by 2-3 weeks. We tracked this across multiple projects and found the recovery phase is where most people give up too early.

  28. I’m sharing this with our content team. We’ve been struggling to explain why «quality over quantity» isn’t just a cliché, and this illustrates it perfectly.

  29. The shift towards «entity-based» indexing is real. Your strategy seems to leverage that by building entity associations rather than just keyword matches. Smart.

  30. The shift towards «entity-based» indexing is real. Your strategy seems to leverage that by building entity associations rather than just keyword matches. Smart.

  31. Great resource. I’ve sent this to a few colleagues who are still stuck in 2015-era SEO tactics. Hopefully, this wakes them up.

  32. This is the missing piece of the puzzle for us. We had the content and the technical SEO, but the off-page signal diversity was lacking. Thanks for the clarity.

  33. Great read. It reminds me of the strategy we deployed last quarter. The focus on foundational stability really pays off when the algorithm shifts. Thanks for compiling this.

  34. This complements the «Entropy» theory perfectly. If you don’t introduce randomness, you’re just painting a target on your back. Glad to see others advocating for smarter engineering.

  35. This aligns with the «Signal Noise» theory we’ve been developing. You need enough noise to mask the signal, but not so much that you lose authority. delicate balance.

  36. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  37. Actually, I have to disagree slightly with the second point. In our testing, we found that over-optimization was less of a factor than pure engagement metrics. It’s interesting to see how different niches react differently.

  38. The analogy of the «immune system» is perfect. You need to build resistance before the virus (update) hits. Too many people react instead of prepare.

  39. This is exactly why we moved away from automated PBNs. The risk/reward ratio just doesn’t make sense anymore compared to what you’re describing.

  40. This is the missing piece of the puzzle for us. We had the content and the technical SEO, but the off-page signal diversity was lacking. Thanks for the clarity.

  41. Does this apply to non-English markets as well? We’re seeing conflicting signals in our EU campaigns compared to what you’ve described here. Would love to hear your thoughts on regional variance.

  42. This is a solid breakdown. One thing I’d add is that the impact of these updates often lags by 2-3 weeks. We tracked this across multiple projects and found the recovery phase is where most people give up too early.

  43. This aligns with the «Signal Noise» theory we’ve been developing. You need enough noise to mask the signal, but not so much that you lose authority. delicate balance.

  44. This is a solid breakdown. One thing I’d add is that the impact of these updates often lags by 2-3 weeks. We tracked this across multiple projects and found the recovery phase is where most people give up too early.

  45. Actually, I have to disagree slightly with the second point. In our testing, we found that over-optimization was less of a factor than pure engagement metrics. It’s interesting to see how different niches react differently.

  46. Is there a specific tool you recommend for tracking the velocity? We’ve been doing it manually but it’s becoming unscalable.

  47. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  48. I’m sharing this with our content team. We’ve been struggling to explain why «quality over quantity» isn’t just a cliché, and this illustrates it perfectly.

  49. I’d argue that the content relevance is even more critical now. We’ve seen perfectly good links get devalued just because the semantic match wasn’t tight enough.

  50. I’d argue that the content relevance is even more critical now. We’ve seen perfectly good links get devalued just because the semantic match wasn’t tight enough.

  51. I’d argue that the content relevance is even more critical now. We’ve seen perfectly good links get devalued just because the semantic match wasn’t tight enough.

  52. I’ve been following this topic for a while, and your analysis on the structural shifts really adds a new perspective. We’ve noticed similar patterns in our internal data at SignalLayer, specifically regarding the volatility timeline.

  53. I’ve been following this topic for a while, and your analysis on the structural shifts really adds a new perspective. We’ve noticed similar patterns in our internal data at SignalLayer, specifically regarding the volatility timeline.

  54. The shift towards «entity-based» indexing is real. Your strategy seems to leverage that by building entity associations rather than just keyword matches. Smart.

  55. Great read. It reminds me of the strategy we deployed last quarter. The focus on foundational stability really pays off when the algorithm shifts. Thanks for compiling this.

  56. Just wanted to say thanks for the detailed case study. It’s rare to see actual data backing up these claims. We’ll be adjusting our Q4 roadmap based on some of these insights.

  57. This complements the «Entropy» theory perfectly. If you don’t introduce randomness, you’re just painting a target on your back. Glad to see others advocating for smarter engineering.

  58. I’m sharing this with our content team. We’ve been struggling to explain why «quality over quantity» isn’t just a cliché, and this illustrates it perfectly.

  59. We’ve been A/B testing this exact hypothesis. Group A (your method) is outperforming Group B by 40% in terms of ranking stability. The data speaks for itself.

  60. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  61. Great read. It reminds me of the strategy we deployed last quarter. The focus on foundational stability really pays off when the algorithm shifts. Thanks for compiling this.

  62. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  63. I’m curious about the sample size for these conclusions. We saw a 15% deviation in our own datasets, but the overall trend aligns with your findings. Good work.

  64. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  65. Brilliant articulation of the problem. The industry has been too focused on metrics like DA/DR instead of actual traffic flow and user behavior.

  66. I’d love to see a follow-up post on how this integrates with social signals. We feel there’s a multiplier effect there that isn’t being fully utilized.

  67. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  68. Does this apply to non-English markets as well? We’re seeing conflicting signals in our EU campaigns compared to what you’ve described here. Would love to hear your thoughts on regional variance.

  69. This is a solid breakdown. One thing I’d add is that the impact of these updates often lags by 2-3 weeks. We tracked this across multiple projects and found the recovery phase is where most people give up too early.

  70. I’m sharing this with our content team. We’ve been struggling to explain why «quality over quantity» isn’t just a cliché, and this illustrates it perfectly.

  71. Brilliant articulation of the problem. The industry has been too focused on metrics like DA/DR instead of actual traffic flow and user behavior.

  72. I bookmarked this for my team. The section on avoiding footprints is crucial. We recently audited a site that got hit exactly because they ignored that principle. Good catch.

  73. Question: Have you tested this approach with expired domains? We’re running some experiments now and the results are… mixed. Your methodology seems safer.

  74. I’m curious about the sample size for these conclusions. We saw a 15% deviation in our own datasets, but the overall trend aligns with your findings. Good work.

  75. I’m skeptical about the timeline you proposed, but I’m willing to test it. If this holds up, it changes how we structure our entire outreach program.

  76. Actually, I have to disagree slightly with the second point. In our testing, we found that over-optimization was less of a factor than pure engagement metrics. It’s interesting to see how different niches react differently.

  77. Great read. It reminds me of the strategy we deployed last quarter. The focus on foundational stability really pays off when the algorithm shifts. Thanks for compiling this.

  78. This aligns with the «Signal Noise» theory we’ve been developing. You need enough noise to mask the signal, but not so much that you lose authority. delicate balance.

  79. Great read. It reminds me of the strategy we deployed last quarter. The focus on foundational stability really pays off when the algorithm shifts. Thanks for compiling this.

  80. The depth here is impressive. Most guides just skim the surface of link velocity, but your point about «natural variance» hits the nail on the head. It’s exactly what we preach to our clients.

  81. Actually, I have to disagree slightly with the second point. In our testing, we found that over-optimization was less of a factor than pure engagement metrics. It’s interesting to see how different niches react differently.

  82. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  83. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  84. The analogy of the «immune system» is perfect. You need to build resistance before the virus (update) hits. Too many people react instead of prepare.

  85. Spot on about the indexing delays. It’s not just about building the link anymore; it’s about the «stickiness» of the placement. We’ve been focusing heavily on that metric lately.

  86. Thanks for the transparency. It’s refreshing to see a strategy that doesn’t rely on black-hat churn and burn. Sustainable growth is the only way forward.

  87. Does this apply to non-English markets as well? We’re seeing conflicting signals in our EU campaigns compared to what you’ve described here. Would love to hear your thoughts on regional variance.

DEJA UNA RESPUESTA

Por favor ingrese su comentario!
Por favor ingrese su nombre aquí